. .. ... U. ”LA U .5 $2 .. . -i - ~ - a.-. :. ..\. .. .. \w . a 14.“. A .V . M Eur. flhs .1 a. a. . 2 . . . .\.\aa 3‘ 53. AV > an». «a W.” i u 0 WM“ QM“ 1.... Q m; .o ,, . E, ,. ,,_ ,2 , fl 0, :5: e. ;_ I 1-.- '5. - . A-Iufl This is to certify that the thesis entitled t ‘ "an EC Oscillator . 'E -..._9r_2*.e.r.1.ted by ”if; f ‘l 5 l-‘coy 'Jo‘rin 'bmollctfl, Jr“. a. o o. ‘ .o‘ ‘- .9: 0.- u.‘ t.‘ a-a-DQ " ' . g h " has been accoptet’toyvards fulfilliLerit" ' 0f thg§gequirement4g__g1,. M . S . fidegree in E169 Em}. git-1.1. [‘3 :3 0‘4 Ho :5 (I) (V ’1’» Ho :3 O“: T ' Major arofflsor Mew 0-169 ‘ c u‘? “ .1 u .‘ . ,. ,.. ‘ I . _ ‘ u' . I '4 "w “1.", x I .~ m—v- o ~— www- -__‘.-—— o . -_—- 1 n v ~-a ._____4_l_.4-__‘ I" II ' ”11“ m. m W“. .m . w T h m D .. u A D A .xw. f x r r. 5.. v _.v.. .C... A. , _ . .a, ......... n -’ . 1 4 x .1 4 . r...r, \ .t . w. t v. \ .w ,ya‘ D I .-. L I .0 .v 1 ) \ cur .. x . x u t . I - ' . I .I (ll! VII. IIHE , ..ltll1‘l.”‘1. Ito. I It '0 I h I'I’I.s Vt O \ | I I ‘ 1 .V - I. c It}. I. .. tll. .‘ I I \l I IlO. Ilr‘ ." I, IK 1! wk .vt ‘. ll.‘ t VI‘ - . ll! \ ;-..-.fi(. IfufW‘ VBV .3 , u a! :r t r a K \ l u I Run- A IVA} « : .1 o IV‘.:.V..1 SHAH/5T1: w...‘h\1u.lfa.l-. d .. u $0.119. .. . . .Uma T / ...\.~fl9n‘; (5" 5.\ LV. oa,l\w\v.. L. ‘\rv..'rIO( uoo.‘ m1... AN RC OSCILLATOR By ROY JOHN SMOLLETT, JR. f A THESIS submitted to the School of Graduate Studies of Michigan State College of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Electrical Engineering 1950 Nature of tie Protl=m . . . . . . Method of Solution . . ~. . . . . The preouency Detcruinlcb Yetue‘k . . ThCr Awflilicl‘ o . o a o o o o o A"plifude Limit Cou+ril . . . . . The Font. 9“;g1; . . . . . . . . Circuit Diagrum Lil ConCLrucLion DEtiilS . ‘ 1 Te-t Results . . . . . . . . . mt — ‘f1 ‘*‘1- .‘ n;ve~Forh OSL;LIOULQNS . . . . . . 13' + - -134‘L1153L o o o o o o o o o o “'3 (D' t (11 ('33 L) (A! Ft] (2) AN RC OSCILLATOR Nature of the problem: The problem undertaken in this thesis is the design, construc- tion, and testing of an electronic sine wave generator or oscillator to cover a frequency spectrum from sub—audio frequencies to radio frequencies. The oscillator is to produce sine waves with little distortion. The frequency calibration should be as nearly exact as can be obtained practically. Method of solution: The vast majority of electronic oscillators consists of the following component parts: a frequency determining network, an amplifier, and a method of limiting the amplitude of oscillations plus necessary associated equipment such as power supplies. This division into parts is somewhat artificial since each is definitely related to the others as indicated in Fig. l. L Frequency i 2 ‘ .__J.n_. 4—,, Determining f A Amplifier ...__.__... Network 1 ‘* Power Supply " ' Block diagram of an electronic oscillator. Fig. l The method of designing such an oscillator is as follows: choose and design a frequency determining network; design the amplifier to provide the transfer characteristics required by the frequency determining network; design the amplitude control network - 2 _ to control the amplifier. And, then design the power supply and any other necessary equipment. Each of these phases of the problem will be treated under its separate heading. The Frequengy Determining Network: The Barkhausen condition for oscillation (Ref. No. l) referring to Fig. 1, requires that the net signal gain through the amplifier and through the frequency determining network will be exactly 1 + jO. An ideal frequency determining network would maintain this condition at only one frequency, and that frequency would be indepen— dent of the transfer characteristics of the amplifier, temperature, and other circuit variables. The transfer characteristics* of such a network plotted as a function of frequency would appear as in Fig. 2. + “ 9(a) x! 3;) 0 ____.JL__. 0 - O .171 W Transfer Characteristics of an Ideal Frequency Determining Network Fig. 2 Such a network cannot exist for it would require no amplifier and would have no losses. However, the concept of an ideal fre- quency determining network has a value as a basis of judging networks that can be realized. For example, if the curves in Fig. 5 were the * Transfer characteristics here referetpwthe ratio of output voltage to input voltage. E out = K(w) E3 E in _ 5 - transfer characteristics of two different frequency determining net- works, other things being equal, network No. 2 would be more desir- able to use in an oscillator than network No. 1, because the. frequency of oscillation would be more nearly independent of phase shift in the amplifier. O 3 Typical Transfer Characteristics of Frequency Determining Network Fig. 5 The "a" (Ref. No. 2) of a network is a much more analytical description of this property. An ideal network would have a Q of infinity. In Fig. 5 network No. 2 would have a greater Q than network No. 1. In general, the greater the Q of the frequency determining network, the more stable the oscillator. The Q of the network is not the only factor to be considered for this application. Some of the other factors are: the number of elements in the network, the time and temperature stability of the elements, the linearity of the elements, and the frequency range poss'ble with the network. Some of the more common frequency determining networks used in oscillators are: quartz crystals, magneto—strictive materials, inductance~capacitance resonant circuits, and resistance—capacitance _ 4 _ circuits. These networks are listed in the order of magnitude of their Q. The magnitude of the Q for a quartz crystal is about 10,000 while a resistance—capacitance network might have a Q of 2. The common frequency range of a crystal controlled oscillator is 100 kc/s to 10 mc/s. A resistance-capacitance oscillator might have a range of frequencies from a fraction of a cycle per second to several megacycles per second. The frequency determining network used in the oscillator to be described in this thesis is a resistance~capacitance network usually referred to as a Wien Bridge. (nef. No. 5) There are several reasons for this choice. It is one of the few practical networks for sub—audio frequencies. The frequency of oscillation is inversely proportional to resistance and capacitance rather than inversely proportional to the square root of these quantities. The oscillations produced are of exceptional purity ' when the network is used with a linear amplifier. The elements of the network are few and simple and are readily available in very stable forms. The network in its simplest form is shown in Fig. 4. Wien Bridge Frequency Determining Network Fig. 4 It is modified slightly when it is used in an oscillator. If the 'l circuit of Fig. 1 is broken at the points 1—2 and the amplifier is replaced by the equivalent circuit of Thevinen's Theorem, (Ref. No. 4) 0’“ o o o "’1 4 the Circuit becomes that of Fig. 5. " 1* ‘ 9 :‘Z-qg . -' f‘ ' f/ ALLA-ALF A 1 II A'A‘.A.A 1:” A P v;:lvv" v ‘" :m (Ir a A- ‘2 j C bl C / fR Foe F., at r ~~ 7 4f 3 l Amplifier Frequency Determining Network Equivalent Circuit of Amplifier and Frequency Determining Network Fig. 5 The Barkhausen condition for oscillation requires that the 0 ED . o o o ratio-e—Q be exactly equal to the gain of the amplifier. It is F12 proposed to make the gain of the amplifier a pure, real number. in analysis of the network with this in view will give the gain requirements of the amplifier and the requirements of the network to produce any desired frequency. Assume no output current and define Zl as the series impedance of R1 and C. Define 22 as tLe parallel impedance of R and Cl' Then :22 f r + E: + 21+ 72 r l + 32 + 21 E12'1 r + Z2 r + 22 This will be a pure, real number if r z: 22 R2 21 Since R2 is a real number, El must also be a real number. r Zn 4 2]. : l + EWCRl 22 = R ij l + jRClW 21 = 1 — “212221001 + j(leC + wRCl) 22 jRCw Then E0C will be a real number at the frequency wo. Where P312 2 R wo RRlCCl=l . And at this frequency 3.1 = fl + .91 . Let 1 = N and zzac E c - '61 = M. Then at W0 , E0” = 1 + N + M if-BZ =1N + M ' E 12 I' 1 Or, if the oscillator is to produce the frequency "0 , the amplifier must have a gain of l + N + M . The frequency determining network must be such that RRlccl == .12 and 32 = N + w . WO r It is extremely difficult to design an amplifier whose gain is a real number over a wide range of frequencies. The effect of phase shift in the amplifier can be most easily understood if the transfer characteristics of the circuit of Fig. 5 are presented as a fUnction of frequency. 31112 = r+Z2 zl men-go g!) Zz= R l+jnfl wo N If .32 2: N + M r .O(w) 12 .__—~L~l—- E . Where C(w) and 9(w) are real. Eoc 1 + N + M V “'2 ‘1 1'" em = 1 + wo2N (“:12 ._ A m ’2 2 WO‘N(R+I‘) l +N + M w“ - E .. —1 (1-3) tan 6(w) = R “ 2 1""! l 2 N w + re + a—(l- l WO‘N YT+R)2 2+§+31+N+M :3) r r R When w ’ wo , 8(w) = tan 0(w) and 9(w) = R .2 y W 1+l'i— ~19 1 .- 2 — N (1—4) V'2N ( D p - 'o 1+-‘-)‘ 2+..+_I: l+N+M no r r R Graphs of these functions are given on the following pages. I- I 11 III l.l‘.‘4 m. on; 3.- 2.- , 9 8" Aw 4“ .< a .a x. 34... ’ Ilc.’J _.J‘I.‘-I J 3.- 2.- . . o o 9. za- L 9.8. a 4. .( .u .3 3. 841 T 10 . 0 1 . 9-..- u 4 .( .w .3 1. {4(1 3-- 2.- . 1 1 _ - 12 - If N = N = l and §'= 1, from these curves it can be seen that a phase shift of only —5° in the amplifier would cause the frequency of oscillation to be about 0.7wO . This would be an excessive error in the frequency calibration. This error can be made a minimum by proper choice of the elements of the frequency determining network. The variables involved are N, M, and E}. This is accomplished by making the phase shift character- istics of the frequency determining network vary as rapidly as possible near w . The curves indicate that this can be accomplished by 0 making 111’ z: 00. This is impossible. However, if .11} >10, the result is very nearly the same. The curves also indicate that M and N should be chosen as large as possible. This is not true, because the phase shift in the ampli- fier is not independent of M and N. The phase shift in a resistance coupled amplifier is given by the relation tan 9a 2 "W“ s . (Ref. No. 5) Where K is the mid—band gain gm of the amplifier, CS is the stray capacity of the circuit, and g7?1 is the transconductance of the tube. In this case, the mid—band gain of the amplifier is required to be 1 + N + M. _ wC£(l + N + M) Therefore, tan 9a = Em Conditions for oscillation require that 9(w) + 9a = O . If 1% =00, this yields -'w he: (1-532) _wCS(l+N+L§) :0 Wo(l + N + it) W’; g m Which gives: 2 2 Cm(l+N+ii 1-“? :_F"0 ( 73) V ng N; This relation is a measure of the theoretical dial calibration error. It is a minimum when N = M = 0.5 . If N = M = l; the ratio of the value of this function to its minimum value is-% . This is no great increase in calibration error, and a choice of M = N = 1 would simplify construction considerably by reducing the number of components of different values. For these reasons choices of N =.M = 1 and«gfl> 10 are nearly r optimum. The approximation that results when considering r = O is not an unreal one. In commercially available oscillators r is not present 0 in most circuits. However, R9 is present in the form of the impe- ') dance seen looking back into the output of the amplifier. Thir is k ordinarily small com "red to R and does not introduce a serious error in the dial calibration. (A serious error here means an error of the order of magnitude of the precision of the dial setting. The circuits that neglect R2 are usually tuned by a variable air conden- ser. The precision of the dial setting is about 1%. If greater precision is required, it is usually necessary to consider R2 .) If the frequency of oscillation is to be varied by a variable condenser, R2 can be neglected for approximate tuning oscillators. R2 can be made a part of R1 for more exact tuning oscillators. In If the frequency is to be var ried by a decade conductance for El and R, r should exist physically for greatest ease of construction and minimum dial calibration error. R2 can be made Quite small if .the last stage of the am flifier is a cathode follower; this vill help keep .3 large. r There are advanV gas and disadvantages to ea ch of these methods of tuning the oscillator. A variable air condenser will give a continuous freuuency spec~ trum over the range of the condenser. The tuning condenser ordinarily can give values of C over a ten to one range. It is possible to get a continuous frequency Spectrum over a wide range of frequencies by including a band switch to change the values of resistance. The precision of dial setting can be made as great as practical by increa 0) ing the dial size and/or adding a worm gear drive to the condenser. The physical size of the variable condenser limits the lowest freuuency of such an oscillator to about 20 c.p.s. The piles sh'ft in the amplifier limits the upper frequency of such an oscilla- tor. . If the tuning unit cor si sts of a decade conducts nce, the freuuency Spectrum of the oscillator is not continuous. However, if the decade conductance is a four dial conductance, the points of the frequency spectrum will be separated by 100 parts in a million. This sepm ation is small enough to be considered continuous for frequencies up to several megacycles. This figure is also the limit of dial calibration of the best variable air condenser. By using a rJo four dial decade conductance, t is possible to set the frequency over four decade bands with a single condenser. This is not possible with the v riable condenser. The lower frequency limit of os*illation with a de caie conducts nce tuni g is a fraction of a eyele per second. It is limited by the maximum resistance value tm t may be placed in tie grid circuit of the an M1 fie er input, and the physical size of satisfactory fixed condens rs. The upper frequency limit is determined by the phase shift in the amplifier and the impedance of the frequency determining network. The decade conductance does offer great ease and precision of the frequency setting. The upger frequency limit is about the same regardless of how the tuning is accomplished. Any phase shift in the amplifier would cause the oscillator to oscillate at a frequency lower than the cali— 1f_ 1 0 phase s: it DJ brated frequency. From the curves it is evident that of but one de gree will cause the oscillator to oscillate ata frequency approximately 95% of the dial setting. The phase shift in the ampli— fier is a direct function of the frequency. The smallest value of the tuning condenser,e with r fivcd Oi variable, should be from 5 to 10 times the stray capacity of the circuit, or 50 to ISO micro-micro-farads. The smallest value of the tuning resistor, either fixed or e out: ut impedan.ce of the a lifiol. :‘i‘ variable, is determined by t When the output stage of the amplifier is a cathode foll_o;er, a lo: outPut impedance stage, the smallest permissible value of tuni e [I . a. resistor is about-% of the cathode follower 1f the 018 torti3n is Em to be kept small. 16 ~ The elimitation ns on the circuit at higher freeuencies place the hi5hest freduency of sa 1 fiactory operation at about one nze5 cycle. In View of the previous discussion, a des ign for a freuuency deterrining net". ore using a three dial conductance is given in Fig. 5. Rb, and Re are all constructed ss shown in Fig. 7. A conluctanc G, endears for R& between terr' nals l l‘. A condac tance EC, an ears for Eh bet een terminals 2-2'. A conductance 3G, appears between terminals S-o' for F . A conductance i9 apnears betusen terminals lwl' for Fb’ a 10 A f“ cond ctan.s ifi a: ears et..oen term nals 4—; for hb , etc. n is similar except they have noninal values over SC. The co.Lim1 tion Of R ’ F"b ’ and R in paral;¢1 tlus form a three dial decade Peasonahle values for these components are: l? ohrs between terrinals l-l' on R,‘ . Renaining values for Rs , <1 A- Eb , and R can be determined as outlined above. C = Euf b , Cd_= .UCEuf, Ce = .u0uou., and Cf = .Qagusuf, This network produces a range of freauencies as listed below: sand Freaucncy Swallest Pesistors Condensers Range Incremental Freuuency l .01- .l cps .Ol cps RC Ca 1 .l — l cps .Ol Pb, RC Ca 1 l - lO cns .31 Pa, Rb, Fc Ca 2 10 ~lTO Cps .l " Cb 3 .33 — l 1rcs. l he 4 l - 13 Res 13 " Cd 5 10 ~13? kcs 30 " Ce 6 1T? —lCOO Ens 1339 " Cf -17- ‘ _ .L 2 - _. Amplifier r Cb [— L 7 Cf % 'l- CC Cd ‘1' 2:0 15 J :1;- Ce 3; I. a _‘ ‘4 ’I ___. \ s_J I \x / :y x r Ra ,i; RD 4: Re Band Witch : Ir Cb- : it .> - ' ’1’ 1! 1V ‘ Ra. AXLE) A’Rc : l I z: | ca I I .____. l l I I Amplifier and Frequency Determining Network \ \f Fig. 6 I 0'. 1'.\2' 5' 4' 5' 6' 7' ' 9' 10' i I I I I I 0. : 2 5 4 5 6 7 8 9 10 l I I q | t l 0.1;2'5 4'5 6' 7' 8' 9'10' Circuit of Ra. Fig. 7 1"" A" : '- ' .. -. W. ~ . dc“..— ° r? . . . TCCLSL—ince 1? L - 1U OLLlsL} )45u1Léd;fl.D r. Tllt ‘iniTLL‘Lvl CLA-E, LLLJ ls 9U L111f0 . :.- ‘ . av‘a. ' V . fi.‘ ~' (“'Ls- .‘ 4"~ v. ‘lfifi The anolifier reduire ants as inpried or Ftnbcd in the preVious freiuency range .Clcps to 1300“cs. The output impedance should be less than 303 Chis. The input impedance should be var" larie; at m .1 least 10 megohn. The distortion should be The shexr; frecuency retuirenents make it necessary to use a direct—coupled a""lificr. (Ref. No. 8) The cathode— ou led amdlifier is such a circuit, and it has reesntly aCILieved considerdtle attention n n for his application. (Pef. no. I,8,3,l$,ll, and 1;) Fig. 8 illus rates the circuit for tii s smilifier. . L + - + 1 E l . 7 cl Er E Actual Circ ?.it of a Cnuhode C oufiil d Inp1i_i Fig. 8 Applying the eduivalent plate circuit t eore ,(Pef. No. 13 <0 this becomes Fi”. L. V -19.. gmlEl +.—-——_ i' , ° ‘ ° ..:. 9 a a * 191'! < Eduivelent Cir uit oi a Cat 03e~Couglc4 eipi -ier 'L'". 0 ILL). u An analysis 09 this circuit gives the citeut voltage, F0 , in terre of th; inuit voltaoes El , a 3 F; , anl the circuit ? re eters. i 1 “l éml'"‘(l + , T ) emirn, F _ a4. ‘4 ’0 1 gi 4‘.l +-: ) ‘1 r0 7 1 L“ ‘ F“ {p +‘l +-l + -l .% +.: *l “ ”l ”“2 r? Ta Z» 5L T“ 1 1 i l J " a 4"; ~ .‘i’. .4. .1. 1 Is- ‘5‘, \VLW + ("1") + + - T": ) $- “";‘ I‘rN r ‘11} .'1 £1” { — Fidn“ - fl 1 l 1 (er +'— +‘: ) 9“"; + ‘_ l‘1. r LJ "~ r M1 2 3‘)”. . ‘h ‘x. 1 + in l + 1 '7 1 N r Ar ..L 1 Em +l + ‘1' L pZLT’H + if +¥ +_ +4?) ul 1'- Z oi. - .q ‘ 41‘, .1 1 J; _} .‘L r1 5 l —2 This eauetion would agnly equally well to the simpler circuit of Fig. 10. n “ ~ _ — {1‘3 0 1 A Secona chivelEFt CirCuit fOr 8 C€tFCCePCovplcd fig: u. .‘. .- ~ ”I ‘ 3 .J 4. -.J L..._....__J q A H + r 3 ILJ A *4 4. “J V on :L‘ A L: 4. NJ V 1‘ J h“ v2. = 3‘3": (1 +% Cm; + {l 1 + l 1 €173 (1 + :1:- -_ — '7. 1 j 3min + 111)..k [efil+%) T) K 1 + 1 + l 1 , = l l. g;:l(l + iv») Eml4 4. l 3 This will be 3 rec l nunle 3r if 7L , Zv , end B are reel. In general, the larger the Quantity BA’" the "reeter will be the reduction of distortion and ou put impedance of the enolifier. This is accorplirhed by ~Loorir; tubes with a large truiscoucuctunce o q o 1 1 o r, l and place resistance, by cLoociLg e rge ZL , a,1 by mCJLRE Ltj>:: . Lin. pis cozbireti n of corc.itirns together t-ith tie coxfitions for a real grin invsriably Make it necessary to provide a grid bias sup131y. This connlicute: the poncr suprly and introduces a 'iuic2el q . -. a 1‘” q _ . .F v N N sources of noise an: um into+ ,2e 0; TCiit Greater tend . 3th for r» the smplifier is also partially assured by chaos ’25 tubes “ithe lerpe trans conductenc e. (Ref. No. 18) It has been suggested tnet tLe elite resistance of a pentode he used for Z». Fef. Lo. 6) Tiis would in neure th t Z» would be <3 (’3 «’3. 124‘ {3 O ":3 {‘1 C+ m 1 (—4- i I»: O verious co idcrot cus—. A conservetive estimete of the tuie para— meters and stray cane city of the circuit gives t2e follohinc values b “ith B O: 55 5% = 3500 u mho (r4 TP 2 4 121L7=31L A = 55 Output Impedance = 400 ohms. Input Impedance - open circuit. Bandwidth — O—SOOkcs. +fz’if‘ " V 5; j; 11.5K ‘110K 31C? "“- kfi“ 6AC7 + ..... 00000 B + t f 6“; 1L W "-1 :2 rd .‘A-A Amplifier Circuit Fig. 12 With E = .55 Gain = 5 Input Impedance - open circuit Output Impedance - 40 ohms Bandwidth O—Smcs. This value of bandwidth is not satisfactory for producing oscillations at 1 mos, but it is just about all that can be obtained with this circuit. Amplitude Limit Control The amplitude of oscillation is determined by the Barkhausen conditions of oscillation. As long as these conditions are satisfied and the amplifier operates in a linear fashion, the amplitude will continue to increase. Obviously, the amplitude cannot continue to increase indefinitely. In many oscillators, amplitude control is accomplished by allowing the oscillations to build up until the ampli- fier is Operating in a non-linear fashion. This generates distortion components which are fed back and amplified again. If the frequency determining network has a high Q, the loop gain for these distortion components is low and a reasonably pure sine wave results. If the Q of the frequency determining network is low, the loop gain may be nearly the same for all frequencies, and the output will suffer con— siderable distortion. A more satisfactory method of amplitude control is to Operate the amplifier in a linear range and control the amplitude by some external non-linear element. A number of circuits have been tested and all of them give excellent results. (Ref. No. 17, 18, 19, 20, 21, 22 and ll) A method suggested by Becker, Green, and Pearson seemed well adapted for this problem. (Ref. No. 25) iThis method uses a thermistor for the nonflinear element. The thermistor is a thermally sensitive resistor with a large, negative, temperature coefficient. (Ref. No. 24) As its temperature increases the resistance decreases. The heating of the thermistor can be accomplished by applying a voltage to its terminals. The q - 2, effect is to produce a non—linear res’stance. Fig. 15 gives a typical characteristic of thermistors. C8 IL] h *L- O 1~+ 4.“.1 Ru. Vifitflr PC" 7"" () The ( Applied Voltage Typical Thermistor Characteristics Fig. 15 This curve is o tained by applying D—C voltages of various values, allon'ng the thermistor to come to a thermal equilibrium and measuring he resistance. It would.be observed eoually well using periodic voltages providing the period of the voltage were short compared with the thermal time constant of the thermistor. This condition is necessary to insure that the thermistor temperature remains constant throughout the voltage cycle. Under this condition the thermistor would appear like an ordinary linear resistor at any given value of applied voltage. If the voltage were increased, and if the thermistor were allowed to come to thermal equilibrium, the thermistor would again appear like an ordinary resistor of smaller value than that observed previously. The application of this property to the present problem can best be understood by considering the gain equation of the amplifier. n The gain or amplifiers with a large amount of negative feedback is given approximately by %- Where B is the ferdhack factor. (Ref. No. 15) The gel: is nearly independent of tote * rimo+ers and circuit constants otrer than the fee dbac d: natuork. Th logical place (D to use the thernistor then, is in the negative feedback netnork. The requirements of this network are that B shall be a constant for cycalic variations of voltage of short period and fixed ampli- tud s, but B shall increase xvith an increase in am; litude. Fig. 14 shows th efecdb ack and amplitude control network and its connection into th as an: lificr circuit. The faeucucc f actor, B, is determined by the voltage divider network composed of the thermistors and the trimmer condenser. The function of the MEGA and WElOA stars is to compensate the network for changes in amoient H- tiiei“n temperature. These thermistors all have about the same te1)w ture coef“icient. However, the WElA thermistor shows a much more rapid change of resistance vith low volt; e tilan do the other two. From J- the curves of Fig. 15 it can be seen that for voltages up to 33 volts across the netvork, the parallel cortination resis.unce of th s 539A and the nTlC A thoifil.’ rs do es not change, while the resistance of };e ”71a th:rni stor till Chang e appreciably. Since the resistance— voltage characteristic of a thermistor is a function of ambient temp~ erature, this combination is virtually independent of temperature. The function of the two neon lamps (NEZ) is to direct couple the two halves of the 85N7 and maintain a fixed potential difference between their cathodes. (Ref. No. 25). The plate resistance of the €SJ7 completes this coupling scheme e. The purpose of this method of -27.. 69“" +zc:v . iEll “K 6A0? V + il +102v 7"K LO Circuit of Amplifier and Agplitude Central Netaork Fig. 14 coupling is to balance the direct current consonant of voltaoe out of the thermistor network and hence increa $0 the sens Ltivity of the am ulLtuce cnrmt 31 The tri immer condenser consensute the "mpli fi1er far phase sLift U) H. U) at hibh freguencies. Its operation a (:1 follows. Due to the strayy capacity of the circuit, the gain of the amplifier becomes a complex numzaer at hibh freouetcies. It is given b) tL3 equation 20 :: A on . 5 El -' I" (RV... IJO. ) l + 3i" + B AB? w T“ J where A and B are defined as before, and w" is the ufper half-power frequency of the amplifier. If B is a real number, the gain is com— plex; and, as pointed out earlier, the phase shift becomes 9} wee )sive at high frequencies. At 500 kcs this phase shift becomes great enough to cause an error of 15% in the frequency calibration. A -28- g .I¥°>J°: 202. Oh Outta-OI 0. _.:3.. complex B could improve this considerably} The efsential elements of the feedback network are shonn in R l C; 92 Elements of the Negative Feedback Netaork E — A —' '5 where D 18 given as From this eguation it can be seen that at any given frenuency the phase shift may be made zero, positive, or negative by afiQusting C. It can not be male zero for all frequencies, honever. For O... i (N f" 'I'vb' values of W £;.l, the phase shift can be made practically zero. The Porer Sunoll- The power supply reguirenents are as follons: {l milliamps at 500 milliamps at 100 volts The 100 volt supply should have an internal imoedance of less than 20 ohns. In additi3n, both of tie :e voltages should be free fr:m L“n and rioole and sLort tLTe flictu ions nit h line voltaoe. These latter reguirements dictate he use of a volta regulated panel supply. (Fef. Ho. 28 The circuit is bLven 11th t11e complete circtu t diagram. It is quite conver ti 3nal with the enception of tLe reference volteoc tubes. Small neon lamps were used for ti is a3plica ion rather than the more conventional "VB" tubes. The current required for these lam3s is abou .8 millianps corparea to £0 millianps for the "VD" tales. “-3 '3‘ (D (J. *3 O (J u (+- H. L) 5 P. U) 4 r3 ‘1 3 f—+ H- 7') F.) .r. 0 CL. 0 ‘3 ((1 O Cirjuit Piathn and Construc ion Detrils: The comb l;te circuit diager is biven on tie foll-oning pages. The corg3nent tarts were mounted on a l7~ by 8» by S-Ln. steel chassis. Every attengt was made to Ice; p stray capacity to a minimum, and the power s1pffl L; was comoletel r=hiel Lei from the rezeinder of the circuit in o15er to Islace hum in the output sigma ThererA1L'Ju—irLLJé CC:' :t'm {pt;‘)?1 ‘.";‘~S COHVC JtiDZ'IL—hl . ‘- J ,.1 3 CL LL L, L -z .3. .L,L "L: .LL. ' Kc; 1..., ,L TLe Le:o til€*COquvab 09c; Ltion dari n ne constLLLtiov r (u aupwuwn passage nopwaaaomo 4 1- Ame -51- um. 1.. W m 4 a a fl f. .v C... N . .J J o u r. g r1T 3 D T I 6 n w m . ,... U «l :7 Wu. 2 up 8 and I d 7.. U. S I. Hun... - 3 TL .3: 1 a .l O 1 bode S I / \\ ) _.||, / \ rl Isl .. \ IT -I I n x / brew Am.HH_ J i is“ —I jK x. MQH 1K 3 YIOMQGN Eututuxeqed Koueubexg bOOH + amHMdHn padonao magnum nmaom newg mom: momm N am 90.. 1f > coH+ 0| J‘ m Jmo fl 0 w. 1: a L: ..... mn+ 1 u" a 1 2-1 m. -a I .K 4 tam :oomw 1 one 100 we; . m 1.1. m d Moqwu1 mo _ ma -1 s a -1 u H _ % 4 . .1 1--- .pH zHL1 Ihmwwn . %1 39H; uopmaoommm mommm fl .0; name L bmnm cam owe cg m m1 Jmoa nwoa -1 ma X 1.11r mm A g . nopfisn ' a... - mfll 1 28 m 1 >m~a awn h owns > can + WMMML mmm 11 11,, mama v, and and 7'7. - Ltd - -m x- .. .-. t.‘ .' . or, v ., 1, . ,-— nuC tie relect- n O; lesistozs ;UI thi FreoleocJ fete; work. Thi netaor< coosictcl of 63 fr uion 1C5;:tk33 with a tolerance of + 17. lere is no cowrereicl source for all the V“1JCQ so it Les decifiefl to sort tLeu _ron 5+;nl'ri 107 composition resistors. The metxofl in which this was carried out use a: fwall:'s: The ZlQS ohm resistor pair of R:i of the frcouency eteimi--nx net— work vac sorted by uswng an imjezcncL tricbe. TFis sub then sol’ered into the selector S?itcl in the oscillator c' CU;U. A c; venient gair of Pm” recrs were also soldered irto the oscillator, and tLe :cillotér mod: to oscilliie. Tfie rcnciri b resistor ;; rs n;ro rel cted b" .kcotirb lesi.93r5 slightly higgo‘ t;;n th; required v: ue and then shurtinr them ti h other resist rs until the correct Lirgajo;s 53 “re (Def. N). 2C, 27) W1« proiet;d on on FClllOLJOpLo It W's necesce"" to use on auxilior; oscillator and make frequent check; against the ,r; gifiol rcr inter plir. Feverel 3i“ ‘re_ of tie congletc: oscillator are {Fouu follotiog a discusrion of tl.e test results. In gencr:l, this orc_ll;tcr terrorms very well. 0 o + q 1 .‘ .1 4. of osc;llation 1° 7 volt - .U volts over its co1piete range 0 -et:ee- i0 C‘" ani L0 kcs the flirtortion is less ing distoritioo No Lthiofi coulfi b The amplitufle f‘rcuiucr‘ c; t r‘4 “a; -r~1a:‘.p ..+.. e observed on an oscilloscope at any frequency above 0.5 cps. No suit ole t* mistor was svelleble xith a tire constant lancer than that of the IFlA. fine: the time cons “st of tLis thermist: r is sbzut 3; second! the arglifiir might be e}: Dected to introduce more disto ition at freiuencies lelét 3.5 C33. A series of oscillogrems is inclossd to illustrate the wove slcge at various frecuencies. The Fico4crc; range of this oscillator 'r 0.31 cps to 530 ”cs. I so: L pc. to make the freqcescy range C.3l C33 t: l mcs ;houever, it res iogossible to congeoscte tie smgliPier for fhsse shift ss-vve 833 k :. Withou uthe fisse s;;-c congcnsation+1he frea¢bl ' rsr'e was 0.31 cos to CO kcs Tie effect of the Leslitcde cortzil circuit is shots in Fig. 17. . 5 5 . .Ildl 4-80.50! \ 802. Oh. Ila-(300 0- .EJHnoHHHomo mac n mgp 20 zoom on ammo Eu .0 mac.. .pse:msow :90 1>.E me emppopowu _UHcp dog 9 opso pr 2H s;g oHpm squunoo on. pmvaamfio opp .omaom mgprnu rpm .wmoo: map :Hmpoo 0p canaroaafiomo egg was hepafiwwoo own some 00.: was Hmw%wamsm an .nmshmoaafiomu «a Howe: has cpumm n so mmqsp mHmE asaamoaawomo mwmgu "muoa one no. pm showm>wfi pepmaawomo n0 n0 -57 mac a pm anommboa Hounaafiomo mac n. no anomopoa nonoaafiomo .17.“ -'n- .xyod hahsmq mm; 8909 02m .maomzm Hmaom mmwpaob swan mnp puoanB ammo was oncomoHHHomo omm Manon 920259 a .wmco:H om was monopmfiu pompwo on» was m.¢ M was msflmomo Hmpnonw one .dzooom a Mo onsmomxm mew» n spa; EHHM oaptfioynozwm no swoop 0903 anSRon momma "mpoz _ $3- moo on no showm>nn MOpwHHHomu mmo com um Showoboa nonmaafiomu mac n no ehommbmm nopwHHHomu _ 59. meg saommbna woo. on Showmpoa Don Ho HopwHHHomo pd. .., Hepmaawo Q moM OOH pm showmbwa 90pmHHHomo was a no eaommpwa pounaawomo _.g)_ .mmooonHHomo .mHnmsnm 1opoes o0 deeper use mHHmume cowpozppmsoo mzHBocw msmab List oP PeCercnces: No. No. No. No. No. No. No. No. No. No. No. No. No. h) U1 (7) 13 11 L. B. Arguinbeu, "Vacuum Tube Circuits," John Wiley l Sons, 1. ’70", L). RtJAQ Cruft Laboratory Staff, "Electronic Circuits and Tubes," McGreW—Hill, p. 32. Cruft Laboratory €ta-f, "Electron ic Circuits anfi Tubes," M'Grew-Hill., p. Ell. Cruft Lahorstory Staff, "Electronic Circuit: an: Tubes," McCrew—Vill, p. 133. Cruft Laboratory 8+5 Ff, "Electronic Circuits and Tuses," McGraw-Hill, p. 345. G. F. Valley, Jr. ani H. Wallmsn, "Vacuum Tube .aolifiers," McCrawnflill, p. 409. ‘vn-- G. C. Sziklai and A. C. Schroeder, "Csthode~Couiled Wide— Band Ampli iers," Proc. I.R.F., Vol. 33, p. 701; Oct., 1915. K. A. Pullin, Jr., "The Cathoce—COUpled Amplifier," Proc. I.R.E., Vol. 3&, p. 402; June, lQiB. P. G. Sulzer, "Freguency and Am :litude Stat ility of t,he Cathode—Coupled Oscillator," PTOC- I.R.E., Vol. 38: p. 5105 May, 1950. M. F. Crothers, "A New B C Oscillator Circuit," 23310 Near, Fadio-Flectroric anireerirg Foi*ion, Vol. 4?, No. 5, Elf-‘3', 1950. P. G. Sulser, "CotLofiewCougled Negative-Resistance Circuit," Proc. I.R.E., Vol. 56, p. 1034; August, lQéB. F. E. Terran, "Padio Engineers' Handbook,“ First Efiitisn, .xvc‘rgl’i—Hlll, p. L 754.. . moi:tau, "Vs HGULW Tibe Circuits, " John Wiley & Cons, . 4'7 ‘ O 4.1.. L. B r 3 L. B. Arguirtau, "Vacuum Tube Circuits,“ John t Chspter VIII. r—y O P .1 C) b) L 7‘) r“) r ) U: h.) D) McCrnw “'11, p. L. B. Prgvixbau, "Va cuum Tube Ci rcui 9," Lohn W11-“ F . ' - w r *3 4 r . ‘r 1" V L. G. Checker? 'rfi F. 0. Wire, "Var 5’1-mF1euh£“c Type Frequenc3. thkilized Oscillatorr," Droc. 1.“. "C' ’3}? p. gut}; {T1TI‘C’ ltZ‘J-LIO L. B. Arguimtau, "An Osci l t v Characteristics," Proc. I.P.F., Vol. 21, p. 14; M .1 Pebenerat on," Proc. I.F.n., ”Vol. 11, . o L“, ) Teacher, "The Brid¢c—Ctaif 11313 Oscillrr+o“," Proc. 1.. an P igrein n and F. m. hailliams, "Theer3 of A27: itufle~FtuLi1ized Oscillgtors," Proc. .P.F., Vol. 7’ 3. .3 January, 1948. Green, ani G. L. Pearson "Proper ies ’ , Uses f Therm15tors,"Trans.A.I.F.F., Vol. 65, p. 711; 4.. O H "I m C- . a. Menu? acturirir , Vol. £2, c. A. W. H. Huggins, "A Ftabilizefl} Icoanute me1ihie Engineering, Vol. 60, p. 437; Septemter, 19:1. Cruft La ”or tory 912 ._f, "Plectro: c Circui s ar1i ucCc“m ~Mil1, p. 571. R. P310 “it Jeru’rv, 1? "Ian PC—Tv;re flurjo- Cigncl Generztor O. ("FIVT‘ .F Conlw nents Open Product De Electricel "IW‘flmjfiflflfl‘lfifl'flfifllfiflfS