AN-238 Technical Information Note I TRANSISTOR MIXER DESIGN USING 2-PORT PARAMETERS Prepared by Ernest Klein Applications Engineering Mixer circuit design may be simpli- fied by the use of small-signal admit- tance parameters. This note describes in detail the effective application of this design technique and the results derived from it. Several design exam: ples are discussed: a mixer circuit converting a 30 MHz RF signal toa 5 MHz IF signal using a 35 MHz L. O. injection frequency with base injection for both the L. . and RF signals: a mixer circuit with a 250 MHz RF and a 300 MHz L. 0. converting to a 50 MHz IF output signal with both base and emitter injection of the L. O. and base injection of the RF signal. The discussion applies to both AGC and non-AGC type transistors. Reprinted by Peratssion of motorola tne(@)}989 AAD nights Reserved. MOTOROLA Semiconductor Products inc. 403 eworonca me.INTRODUCTION ‘Mixer stages perform thefunction of frequency conver sion ina signal chain, In receivers, mixers commonly transform the incoming signal frequency to an interme- diate frequency, or transform one intermediate frequency to another in multiple conversion receivers. In trans~ mitters, the increasing use of the suppressed carrier Single sideband (S55) mode of communication has brought about much wider usage of mixers since frequency multi- plier stages cannot be used in an SSB signal chain, Design of small signal high frequency amplifiers has reached a Aigh degree of sophistication through two-port parameter analysis, These techniques have been well documented. 113 However, literature on the theoretical and practical design aspects of the solid-state mixer is less plentifal and often less precise than the literature on amplifier design. This report presents a design approach which is based on the popular two- port parameter amplifierdesign meth- od. A theoretical discussion and design method are de- scribed together with detailed descriptions of specific mixer circuits DESIGN CONSIDERATIONS Inthedesignof mixers, the following items are of con siderable importance in obtaining a good overall eireult. (2) Frequencies () stabiuiey (©) Gain (@) Network Design (@) Local Osetllator Injection (0 Device Selection Each of these items are important in the clreult design considerations for the following reason: (2) Frequencies: Mixing action is done by means of a non-linear device, therefore, many different frequency components will be present in the output of the circuit. These {requency components may be categorized as: 1. Spurious Mixer Products - All frequency compo- sins tn the Supt ol tice othe: than the Sealed ‘sum or difference output component, 2, "Crossovers" or “Birdies” ~ Undesired mixer fre- ‘quency components wnich fall within the mixer output passband. 3. Intermodulation ~ Distortion Products - A special lass of spurious mixer products falling within the mixer output passband, and resulting from interac~ ‘Hon between signal components fed into the mixer. ‘A problem with any mixer is spurious output signals. In addition to the obvious outputs of the local oscillator frequency, the input RF signal frequency, and the unde- ‘sired sum or difference frequency, many other spurious ‘output signals may be present. Many of these additional ‘spurious outputs are due to third and higher order dis tortion characteristics which the non-linear device ex- hibits in addition to the second order distortion utilized bby such a device to produce the desired mixing action. Spurious output signals at both the desired output {re- quency, (classified as "Crossovers" or "Birdies") and Other fréquencies may be present. The “olf” frequency signals may be attenuated by tuned cizcuits following the mixer, but the only recourse to spurious outputs at the desired output frequency may be an entirely ditferent ribeing scheme, Literature on the selection of mixing frequencies to rinimize spurious outputs is available, rence, the sub- ject will not be covered further here, 2 ‘This report assumes that the designer has already se- lected suitable mixing frequencies and is ready to pro- ceed with the actual mixer design. (®) stabinity: ‘The stability problem may be locked at basically from 1 two-port standpoint. In the design of high [requency amplifiers, the Linvill stability factor may be used to determine the potential stability of the transistor, The Linvill stability factor C is computed from the foliowing ‘expression: ¢ olan tal 2641 62a ~ Re Wy2 Yay) XE tn tess than 1, the transistor 1s unconditionally szable, ie G is greser tan y the transistor is pote aly inate ‘The C factor sa test for stabllity under a hypothetical worst case condition; that is with both input and output terminals of the transistor open circuited, With no ex- ternal feedback, an unconditionally stable transistor will ‘ot oscillate under any combination of load and source admittances. Ifa transistor is potentially unstable, then ‘with certain source and load admittances oscillations will ‘The C factor is used todetermine the potential stability of the transistor, but itis often desirable to compute the relative stability of actual circuits. Stem has defined a stabillty factor k for this purpose. The k factor Is sim- Har to the C factor except that ittakes into account finite source and load admittances connected to the transistor, ‘The expression for k is: 21146, y+) Paz Yaad * Fe Oz Yaa) If kis less than 1 the circuit will be potentially un- ‘stable while if kis greater than I the circuit willbe stable, For the collector circult of the mixer, a stability fac- tor could be calculated at the output frequency, and for the base circuit a stability factor could be calculated at the input frequency. The greatest danger of oscillations cceurs at the output frequency since the impedance level at the collector is higher. Circuit diagrams are included as 4 means of illustrating typical semiconductor applications, consequently, complete in- formation sufficient for construction purposes, is not necessarily given. The information in this application note has been carefully checked, and is believed to be entirely reliable. However, no responsibility 1s assumed for inaccuracies Furthermore, such information does not convey to the purchaser of the semiconductor devices described any license under the patent rights of Motorola Inc. or others. 404From the k factor, ve can see the desirability of hav- ing low source = impedance and load - impedance olf- frequency, and therefore assured stability. If the output port presents a low impedance or short circuit at the In- ‘put frequency, and the Input port, presents 2 low imped- nce or short ctreuit at the output frequency, then oseil- lations will not occur. In case it 1s impossible to meet the above stability conditions, as calculated from the k factor, an IF trap circuit could be used at the input of the two-port network ‘and an BF trap circuit atthe output of the two-port net= work, (©) Gains Conversion gain is defined as TE Bower out RF Power in Depending uponthe frequencies of operation, the mixer conversion gain will vary as does the gain of any ampli- Her designed for operation at the ditierent frequencies, ‘The gain, however, will be comparable to that obtained nan amplifier designed for the same IF frequency of operations (@) Network Design: ‘The primary considerations in the design of the input land output networks are conversion gain, stability, and attenuation af off frequency spurious output signals, (©) LO. Injection ‘There are basically two methods for injecting the L.O. (Local Oseilator) signal, either base injection or emitter Injection. Of these two schemes, base injection some- times provides greater stability. Less stability problems are encountered, especially in. the VHF and UNF fre- quency ranges, with base injection than emitter injection 4nd these problems will be explained in a later sectionon Design Theory. (Device Selection: 5 Since the transistor is a non-linear device, any tran- sistor could be used as a mixer. However, certain char acteristics make some transistors more desirable than others. A. Frequenctes: The device must be capable of opera- ‘at the input, output and local oscillator frequen- Gain: The device gain wil be within 3 4B when used as a mixer as opposed to an amplifier designed at the output frequency in the unneutralized condition. ‘Therefore, thedevice must be capable af the desired fain at the output frequency. bbe desirable though not necessary as explained in the ‘section on stability. 4. Depending on the application, either an AGC or a non-AGC device may be chosen. 5. A device with a low input capacity, Cin, provides for fasier impedance matchin DESIGN THEORY ‘The basic design theory of a mixer circuit will now be explained in terms of equivalent circuits with emphases fon impedance matching techaiques af both the input and output of the transistor to the respective Networks, ‘The basic mixer circuit can be explained in terms of the following equivalent circuits: For maximum ciresit stability with the input ctreut at the RF. frequency, we ‘may consider the output to be a short circut, SURE aac EUNALENT cncuiTs nea witm GUTPUT From the equivalent circuit a vy Nie vyr0 For the output circuit atthe LF. frequency, and max- imum circuit stability, we may consider the laput cir cuit to be a short circult at the LF. frequency. ‘RQURE 2~ EQUWALENT CRCUIT: MIRER WITH INPUT SHORT ERGRE 5 Faun iER WITH INPUT Prom the equvalnt ctr E Yo00 77 tly 00 Inne move woeqivalentcreits yagi the common cater short creat np untaney 280 ne Geto mon emitter short circuit output admittance of the tran= sistor. Yin" Yue Yout * Ta20 a — = — 5 > | ourprr networs | Yun Yin | NETWORK | Youp Yout | NETWORK [AGUFE 3 — NETWORK INTERCONNECTIONS 405Under these conditions in the network shown in Figure 3, Yin represents the input admittance of the transistor and 1g equal to the small signal common emitter input admittance ¥i1¢ at the input frequency. out represents the output admittance of the transistor and ig equal to the ‘small signal common emitter output admittance Y22e at the output frequency. For maximum conversion gain we know that the input network should be conjugately matched to the transistor Input admittance. Also, the output network should be conjugately matched to the transistor output admittance, ‘Therefore, in Figure 3, Yiq* is the conjugate admittance Of Yipy and Yout" 18 the ‘onjugate admittance of Yout. Inthe design of 2 mixer circuit using base injection {or both input and LO. signals, as long as the base does not have sulficient drive from the L.O, signal to move operation out of the small signal region, then the small signal common emitter short circuit input admittance will be the design criteria for the source admittance, AS the LO, level is increased, however, the transistor Is driven harder into conduction. ‘The Input admittance of the transistor changes to a large signal input admittance, this new large signal input admittance then becomes the design eriteria for the source admittance. Lr Load 'AGURE 4 — BASE WJECTION OF BOTH R. AND LG SGHALS ‘Assuming operation in the small signal region, (refer toFigure4)the source at the input frequency should con- stitute Y1t¢ conjugate and at the output frequency a short ‘ircult,aiso the load should appear as 22e conjugate at the output frequency and a short circuit at the input fre- quency. In Figure, emitter injection is used for the L.0. sige ral with base injection for the input signal, Ifthe L.O. were tobe injectedat he emitter instead of the base, then the emitter would still have to be at RF. ground for the input frequency but at a alfferent impedance level {or the L.O. signal, This requires more complex network de- sign for the L. 0. injection network. Of the two methods discussed forthe L.O. injection, base injection and emit- ter injection, the base injection method Is easier to work with from a stability standpoint, In the previous discus~ Sion the emitter has been at RLF. ground for the input frequency with both injection schemes, In the case of emitter injection, the emitter cannot be bypassed at the LO. frequency. This, therefore, requires a network presenting to the emitter a short circuit atthe input fre~ Queney and an impedance which is the conjugate of the L.0. source impedance at the LO, frequency. This is 2 duftieult design problem ifthe LO. and input frequen- cles are close together, as they are in the VHF or URF frequency range. MIXER CIRCUITS ‘The design techniques and requirements discussed in ‘thepreceding sections will now be applied to various cir- cult examples. These examples are of a general nature {or use over the VHF and UEF frequency ranges. They havenot beendesigned for any particular application oth- ‘er than to illustrate design theory and design variations. ‘Three baste circuits are discussed: (1) a mixer con- vetting a 20 Miz RF signal to a § Mile IF signal using a 35 Milz L.. injection frequency with base injection for both the L.O, and the RF signals using a 2N2Z21A Mo- torola transistor, (2)a mixer converting a 250 MHz RF anda 300 Miz 1.0, toa $0 Miz LF, frequency using bot base injection of the RF and L. O. trequenctes for 2 Motorola MMI941 non-AGC transistor. (9) a mixer cir~ ccultof the same scheme as the above but using a Moto — ola 2N3308 AGC transistor, These various circuits were Gesigned, and coaversion gain measured as 2 function of various parameters. ‘The results are given in this sec Hon. ‘The following {4 a mixer design using & Motorola, 2N2221A transistor and converting 2 30 Miz signal to a 5 Miz LF, signak First obtain admittance parameters at te desired frequency. For the Motorola 2N2221A, Gg=2mA, Yog = 10 valis), Yate (6-25 + 9.5) mmbos (20 aie) Yogg * (0.027 + } 0,28) mmhos (5 Miz) Consequentiy, the design eriteria for the input network would be Uvgig + [3002 | 10.857] And for the output network Uygq + [37 koll -9.0 2] ‘Assume thatthe load the output network must be matched toils $0 ohms. Also assume the network to be af the low- pase filter type such that the higher {requencies will be Attenuated, such as the one shown below. SOURCE RGURE 4 — BASE WIECTION OF RF. SIGNAL INO EMITTER NUECTION Sauna GRA ANG BATE NOTE: "+", refers to conjugate form, 406 Ly i L ‘The cireult Q may be defined as follows for the series2 a fee ar BS Where Gp is the real part of the parallel admittance and Rg is the series resistance to be transformed to, repre- sented by the 802 load. i fate a0 tie m3 Re Ld Fr % we may compute the required inductance Ls. J Ly = RQ, = OO 6 43.5 a eae (6.0 x 105 (6,28) ‘The reflected parallel inductance, Lp, may be com- poted from a? 1, >t, (1+179,") fm the above case Lp = Ls * $1.5 ll The required cafucltngs fr reecnance at SMa maybe called fom 1 t9a5 Je where L = 43,5uB. In this case Cp = 25 pF. From the y-parameter equivalent of the transistor the imaginary art Tepresents 9.0 pF of capacitance. Therefore, an ‘ditional capacitance f(25.0-9.0) pF = 16pF isneeded, For the input network, again assuming a 509 termina~ Lon, let a transforming network be used as illustrated Si o—+— s Rg= 50a which ts equivalent to: ——— using the relationships as described above: VB ea if 1 ve," (8.0% 10") (6.28) (1. 47) (50) = SEE. 49,5 0F res raual ter ‘The total capacitance which must be tuned out at the base is equal fo Cp + Cin of the transistor, or 49.5 + 50,5 = 100 pF, Aninductance value of 0.22 uH was selected to permit the addition of a trimmer capacitor from base to ground, ‘The complete circuit is as shown below (Figure 8) ‘The curve in Figure 7 gives conversion gain as 2 fune~ tion af L.O. input levels for different collector currents. Since the y parameters are a function of collector cur- Fent a new set of parameters were used to design the matching networks for the different collector currents, ‘The LO. voltage is measured at point (A) in Figure § above. Inallfollowing circuits the L.O. voltage is meas- ‘ured at this point, ‘The following are results obtained using an MM(194i ‘Motorola silicon NPN transistor ina 250 MHz mixer con- verting to a $0 MHz output frequency. The eireuit, Fig- lure 9, shows the transistor and matching networks for both the input and output. Figures 10 through 14 are curves of data obtained using the circult shown in Fig urea, anz221A, soa 500 AGURE:6 — A 30 wie RF TO 5 Mis IF MER 40750 40 : oa pf 3» a Zz a LL 3 mA - & ‘2N2221A MIXER (NON - AGC) 2 (G5MEiz, 30MG2 — SMzz) z eens 5 8 “10 16 02 0.4 06 08 Lo L2 La 1 1.0. INPUT (VOLTS) FIGURE 7— CONVERSION GAIN VS LO. INPUT LEVEL USING MOTOROLA 2422214 TRANSISTOR o—1 [AGURE 8 — CIRCUIT GINGRAM OF 283308 MIXER 300, 250 Mir— 50 Mix o—I [AGURE 9 — CIRCUIT OUGRAM OF M941 MIXER 300,250 MHz —~ 50 MN 408Figure 10 shows the relation existing between conver~ ston gain and source admittance for changing values of RF Source admittance, For the transistor being used 7136 o.4e] 15 mmbos, and 1/Yite = [1199 || 7.3pF). Curve A shows the variation in conversion gain as the Fealpartof the source admittance is varied while holding the (maginary constant at -7.9 pF. Curve B shows the ‘variation in conversion gain as the imaginary part of the Source admittance ie varied while holding the réal part constant at 1192. This confirms the fact that maximum conversion gain occurs very near the point where the source admittance presents tothe transistor the conjugate Of the transistor input admittance, Figure 11 gives conversion gain as a function of L. 0. Injection capacitor size for a constant L.O. injection voltage at 300 MHz. 0 ‘1941 MOXER NON AGC ‘S00MEEs, 250MEe —~ SOME Lamy’ F. INPUT 3s f i For TRassisToR = (18085 [7.3997 z Zz % 7 3 + (190HMsS [| - 7.39F) ; aun g 2 = 5 | ry & 2. ri 7 CONSTANT -1.39F go ate CONTANT = T.88P gz-hys= ConstaNT iteoms \ 7 RESISTANCE VARIED CAPACITANCE VARIED 10 ° 30 709 150 200 350 300 350 00 aelyg RESISTANCE-omns “is “10 5 ° 8 “10 8 Pours 19 ou as aruneron oF source THY! CAPACITANCE-PE Rintoaaee Gerakan 38 wanci941 MEER 7 00M, 2506it2 — SoM: Lonv RF. INPUT 0.8 VOLTS 1.0, INPUT = 3 3 z § 2 z g Bos 8 8 10 5 ° oo5 1 Ls 2 hes 8 35 s&s 5 1.0. CAPACITANCE (pF) FAGURE 12 — GAN AS A FUNCTION OF Lo. INPUT CAPACITANCE 409Figure 12 is a curve of the measured real part of the source admittance for maximum coaversion gain for dil- Ierentlevels of L.O. input voltage. Figure 13 is 4 carve of the measured tmagiaary part of the source admittance for maximum conversion gain for differentlevels of L. 0. Input voltage. These indicate that the input admittane of the transistor changes as the L.O. drive is increased And, therefore, out of the small signal region of opera ton of the transistor. Figure 14 is the difference in conversion gain for dif- ferent LO, input levels when the source admittance 1s matched in’ conjugate form to the transistor parameters and when the source admittance ig tuned for maximum ‘conversion gain. As can be seen from Figure id maxi- mum conversion gain occurred at an L.O, level such that operation was stil inthe small signal region of op- ‘eration of the device. Figure 15 gives conversion gain as a function of LO. input voltage for different transistor currents, As can be seen at the higher curreat levels, the conversion gain changes somewhat over the range of L.O. input voltage While at the lower currents the conversion gain remains nearly constant. ‘The same design procedure was used to design a mix or circuit at 250.200 tai with a 50 MHz output frequen= ey using a Motorola 2N3206 silicon PNP transistor. This Is an AGC device as opposed to the MMI941 which is 2 non-AGC device. This transistor and its associated in- [Put and output matehing networks are shown in Figure 6 Figure 18 shows that the elfect of AGC action takes place between two and four mA af collector current, ‘The preceding circults are based on basd injection of both the RF and L.0. signals. ‘The emitter was always bypassed and, therefore, at RF ground. Foremitter infection of the 1.0, it was found that ete cuit oscillations were harder to control than in the case of base injection. This would indicate thatthe Input RE network was not appearing as a short circuit to the IF frequency. It was also found that the elfect of RF sig- nal frequency and L.O. level on conversion gain are more pronounced in this case than in the ease of base injection of the LO. Also, optimum L.O. shifts upward with ine creased emitter current. a T T NPN ‘MM 1941 MIXER ‘NON AGC 3o0nis, 250ne — sone Lemv RP. INPUT g 1 | 1 ! a Fey] MATCHED To TRANGETOR 2 3 100 ; —L i aed FoR Mamie eowenson cans 20 az ae iF Lo. mur (vous) RGURE 12 — Orr rr NETWORK RESISTANCE FOR MASIMUM SU dan aa 10 NPN T NON Ace 1m ( ¥) MArcHED 70 TRANSISTOR ° = a 8 3 hM_] E go4 z 3Ot 1901 seER 3 ‘300MH2, 250Miz—~SOME2 | — Lomv a. INPUT 7 Tm ( ¥,) FOR MAXIMUM CONVERSION GADT ° | 0.7, a8 0.8 0.8 10 La Le 16 j 1.0. INPUT (VOLTS) RQURE 13. OPMMUM INPUT NETWORK INDUCTANCE FOR MARU SGNVERSION GAN AT VARIOUS EO. LENELS 410CONVERSION GAIN (4p) CONVERSION GAIN (4B) ‘CONVERSION GAIN (4B) 0 30 20 10 =10 0 -10 40 -10 NPN MM 1941 MEER NON AGE ‘300MH2, 25032 — SOMES 1.8 mV RF. INPUT MAXIMUM CONVERSION GAIN | IMPEDANCE MATCHED CONVERSION GAIN t ° 0.2 4 0.8 0.8 u 14 1s 1.0, INPUT (voLTS) AGURE 14 ~ IMPEDANCE MATCHED CONVERSION GAIN NPN MM 1941 DOXER S00MHs, 250MH2—~ SOME Now age 6 mA 4.8 mVR.F, INPUT, 4 mA nu i = | 0.5 mA 2 mA ima ° 2 a4 0.8 0.8 4 6 1.0. INPUT (VOLTS) FAGURE 15 — CONVERSION GAIN FOR DIFFERENT CURRENTS. PNP 4mA 2 mA FI {| 2N3308 MIXER ‘AGC DEVICE S00MHs, 250MH2—- SOMES amv & F. INPUT Va ° 0.2 04 0.8 0.8 Lo 2 1A 18 1.0. INPUT (VOLTS) FPGURE 16 —~ CONVERSION GAIN FOR DIFFERENT CURRENTS 411CONCLUSIONS: ‘The design examples in the preceding section of this application note showed in detail the design technique for amixer circuit using base injection for both L.O. and [RF signal inputs. The results of the mixer design are consistent with the theory explained in the preceding sec- Hon, Maximum conversion gaindid occur when the source admittance was the conjugate of the transistor input ad- mittance and the load admittance the conjugate of the transistor output admittance. The best stability was ob- tained when the source admittance presented a short clr- cuit to the IF frequency. The conversion gain was also found to be comparable to that attained for an amplifier designed at the IF frequency. SUMMARY OF TRANSISTOR MIXER DESIGN PROCEDURE, In addition to selecting a circuit at the input to match the transistor input admittance and a circuit at the output to match the transistor output admittance, other factors must be considered, A suitable network should be chosen toreduce spurious or undesired frequency components at the output due to the non-linear character of the mixer Itself. "At the input, the network may be used to obtain a desired selectivity if other frequency components are available other than the desired input signal. From the curves on conversion gain as a function of ‘Local Oscillator input level, the following comments are evident. The level of operation can be left up to the n= dividual designer, but is relatively unimportant as far a3 conversion gain goes, a5 long as the level does not ex- eed the region of small signal operation. ‘A suitable coupling capacitor in the VHF frequency range was found to be about 1 pl. This value, inthe case of Base injection, does not considerably alter the tran- sistor input admittance appreciably from the theoretical value of 749 Without Local Oscillator injectioa, In choosing a bias point of operation for the transistor, and neglecting noise consideration in the device and as Sociated circuitry, a good design criteria would be the Same as that used for small signal amplifier design, ‘The cholce of either aforvard AGC or a non-AGC tran- sistor isafunction of what the individual designer desires. GLOSSARY © = Linvit's stabiity factor k © Stern's stability factor = Real part of the source admittance ‘Real part of the load admittance Imaginary part of the source admittance = Imaginary part of the load admittance = Real part of 94, Su qq + Real part of Yop BIBLIOGRAPHY 1, “RF Small Signal Design Using 2-Port Parameters," Motorola Semiconductor Products, Tne. , Application Note 215A, 2 tSingle Sideband Principles_and Circuits," by £. Pappenfus, Warren Bruene, =. Schoenike, McGraw. Hu, 1964 3. "Stability and Power, ‘ned Tes : TRE, March Plifiers,"" by Arthur P. Stern, Proc. 1987, 4m ircult Engineering,” by R. F. Shea, Toha Wiley and Sons, Me., 1957" MOTOROLA Semiconductor Products Inc. 412