AN INSTRUMENT FOR RECORDING THE FREQUENCY DRIFT OF AN OSCILLATOR

Transcription

1 JANUARY AN NSTRUMENT FOR RECORDNG THE FREQUENCY DRFT OF AN OSCLLATOR by W. W. BOELENS : " A quarz oscillaor has a frequency ha is sufficienly consan for deermining hefrequency drif of an ordinary valve oscillaor due o changes in emperaure, for which purpose i is in fac he mos suiable oscillaor. This frequency drif, for insance ha of an oscillaor in a superheerodyne rcr.eiv6r, should be measured, however, in more han one posiion ofhe uning capacior. The problem is how o derive from one quarz crysal a number of reference frequencies spread more or less regularly over he frequency range of he oscillaor o be measured. Wih he frequency-drif meer described here his problem has been solved by generaing in a number of frequency-muliplying sages he weny-fourh muliple of he crysal frequency and by an arifice deriving herefrom nine oher muliples (he 19h up o and including he 28h). is a well-known phenomenon ha afer a valve oscillaor has been swiched on is frequency ends o drif. This is due o he fac ha when an oscillaor is swiched on various sources of hea are brough ino play (valves, resisors, ransformers), so ha he emperaure and hus also he dimensions and hence he self-inducance or capaciance of he elemens in he oscillaing circui undergo a gradual change. n pracice i ofen akes some hours before a nearly sable condiion is reached. f he oscillaor in quesion is par of a superheerodyne receiver and no adequae seps have been aken o counerac his phenomenon hen, as a resul of his frequency drif, he uning of he apparaus may have o be readjused several imes o ensure saisfacory recepion of a paricular saion. This is especially imporan in he recepion of waves shorer han 10 m. Seps aken o preven' his may consis in he applicaion of. hermal insulaors around he sensiive pars, he provision of venilaion, he use of circui elemens having a low emperaure coefficien, ec. Thus, afer having applied one or more of hese measures in an experimenal apparaus, he designer of a receiving se is faced wih he ask of asceraining wheher he defec, in quesion has been sufficienly remedied. This he does by measuring, wih reference o a reference frequency, he frequency drif Lfo as a funcion of he ime afer swiching on, and his involves ~ series of measuremens ha may have o be exended over a number of hours. As a rule he firs resul will no be saisfacory; a suspeced elemen may have o be replaced and he measuremens hen repeaed all over again.. does no suffice simply o measure he final value of he drif, because i may have previously reached much higher values. Thus he drif (and by ha we mean here he absolue value of he frequency variaion) as a funcion of ime may show a maximum somewhere on he curve (fig. ). This is due o he fac ha he oscillaing circui comprises a number of elemens (inducance of he coil and of he wiring, capaciance of he uning capacior, of he wiring and of he valve) some of Ll~ o 'KJ' 20' JO' 40' 50'!.!:.._ Fig. 1. Example of he variaion of he frequency drif Lfo of an oscillaor as a funcion of he ime afer swiching on. n he case drawn here is a maximum. which may show a posiive and ohers a negaive emperaure coefficien of he inducance or of he capaciance as he case may be. n course of ime here may be a cerain degree of compensaion which does no exis a he beginning. canno be prediced a wha sage he maximum may occur afer swiching on because of he grea differences in he rae a which he various componens reach heir final emperaure: he valves and hose pars soldered ono he valve sockes may aain heir final emperaure in 10 or 15 minues, whereas he pars mouned on he chassis may ake a much longer ime. Consequenly i is necessary o ake he measure-

2 ---: ~- -..,- 194 PHLPS TECHNCAL REVEW VOL. 12, No. 7 mens a sufficienly shor inervals of ime o ensure ha any maximum occurringin he frequency drif does no escape noice. Firs hey should he aken, for insance, every hiry seconds, hen every minue, afer a quarer of an hour every five minues and afer an hour every fifeen minues. These inervals are no long enough o permi he observer o occupy himself seriously wih any oher work in heween, so ha for ha reason alone such drif measuremens cos a grea deal of ime. Furhermore, he ploing of one drif curve alone does no suffice, i heing necessary o plo a curve for several posiions of he uning capacior, since he various capaciances in he oscillaing circui vary according o differen funcion.s of ime, so ha measuremens have o be aken for differen raios of he componen capaciances. Very much ime can herefore he saved by employing a recording insrumen which, once i is sared, auomaically plos he drif curve, Below a descripion will he given of such a recording drif meer in a form specially designed for esing F.M. radio receivers, he effecive frequency range of which exends from 88 o 108 Mc/s. Principle of he drif meer F.M. receivers usually work wih an inermediae frequency of 10.7 Mc/s, hough lower values may also occur. depends upon local condiions (near-by ransmiers working on adjacen frequency channels) wheher he designer finds i beer o choose an oscillaor frequency fo for he receiver ha is higher or lower han he frequency of he saions o he received, wih he resul ha in pracice boh soluions occur. Therefore, in order o he of use for as many ypes of receivers as possible, he drif meer mus be designed so as o cover a range exending from = 77.3 Mc/s o = Mc/s. The maximum drif ha can be oleraed in his frequency range varies from 25 o 50 kc/s, according o he qualiy of he receiver, and hus he drif does no amoun o more han abou 0.05%, which is far oo small o he measured direcly wih sufficien accuracy. The bea frequency of Jo is herefore formed wih a neighbouring reference frequency fr; his bea frequency shows a relaively much larger and hus more easily measurable drif. The reference frequency fr canno be aken from a valve oscillaor wih an ordinary oscillaing circui because fr would hen iself show variaions of he same order as he drif of Jo. For generaing he reference frequency fr one mus herefore use a cry sal- con r ole d oscillaor. The difficuly ha a crysal gives only one fixed value of fr, whereas Jo is o be measured for a numb er of values (in he range from 77.3 o Mc/s), can be me hy choosing a crysal frequency fe much lower han fo, say fe ~ 4 Mc/s, and producing, hy n-fold frequency muliplicaion, a sufficien number of muliples of fe which lie wihin his range and hus can serve as reference frequencies Jr. Wih fe = Mc/s (he value acually used) en useful reference frequencies are ohained: Jr ~ 80.4, 84.6, Mc/s, a sufficien number of which will always lie in he range covered hy he oscillaor, no maer wheher he oscillaor frequency is higher or lower han he signal frequency, and for all inermediae frequencies up o well over 10 Mc/s. To ge hese en reference frequencies i is necessary ha n can assume he values 19, 20, Now his is easier said han done. nhe frequency range _specified a 20-fold muliplicaion wih such a seleciviy ha he adjacen muliples do no inerfere canno he hrough ahou in one single muliplying sage 1). is rue ha good resuls could be reached hy employing more han one sage, each wih a small muliplicaion facor, hu of course his only helps for hose values of n which can he resolved ino small facors, such as 24 = 4 X 3 X 2, and 27 = 3 X 3 X 3, so ha he prohlem of geing n = 19, 22, 23, ec. canno he solved in his way. To overcome his difficuly he following arifice has been employed., As indicaed in he hlock diagram of fig. 2, from he crysal frequency Je = Mc/s a 24 imes, as high reference frequency of Mc/s is ohained in hree sages. Boh a volage of his frequency 24Je and he normal volage ofhe oscillaor frequency Jo are applied o he mixing valve M 1 of he receiver under es. The volage a he oupu of his valve has he much lower hea frequency 124 fe-{nl. Afer amplificaion, accompanied by he removal of undesired combinaion frequencies hy means of a uned circui, his oupu volage is applied o he inpu 1 of a second mixing valve M 2 forming par of he drif meer. To he inpu 2 of he laer valve a volage is applied having he crysal frequency fe iself and of such an ampliude and hias 'ha he valve is rendered conducive 1) n principle he series of harmonics obained wihou his seleciviy are jus suiable for he purpose, bu he cornponens are so small in ampliude ha considerable amplificaion is necessary, and a frequencies of abou 100 Mc's his gives rise o grea difficulies. is furhermore unknown which of he harmonics is serving as a reference frequency.

3 JANUARY 1951 RECORDNG FREQUENCY-DRFT-METER 195 only for a shor ime a he posiive peaks of he volage a he inpu 2. This has he same effec as if his.were a pulsaory volage wih he repeiion frequency fe. Consequenly he valve shows a conversion conducance - he raio of he A.C. oupu curren wih he bea frequency o he A.C. inpu volage a 1 - which, as a funcion of he c Fig. 2. Block diagram showing he principle of he drif meer. R receiver wih oscillaor 0 (frequency fo) he drif curve of which is o he recorded, and mixing valve M l C quarz oscillaor (frequency fc), D frequency mulipliers (24 X), Ma second mixing valve, Al and Aa amplifiers, L limier, FD frequency deecor, pa recording moving-coil meer wih scale calibraed in frequencies. The pars wihin he recangle F helong o he drif meer. frequency a inpu 1, is approximaely equal for frequencies fe and no oo large muliples of fe. Thus, in he main, he oupu volage of he valve conains no only componens wih he frequencies 124 fe - fa ± fel bu also componens wih 124 fe - fa ± k fel = 1(24 ± k) fe - fol, where k = 1, 2, 3 and so on. The shape of he pulse is such ha he componens wih k = are presen wih he required ampliude. The effec is herefore as if he reference frequencies 19fe... 23fe and 25fe... 28fe were all presen simulaneously (he omied erm 24fe will be deal wih laer; 29fc is no in he desired range and can be lef ou of consideraion). Afer fa has been brough sufficienly near one of he above reference frequencies by urning he uning knob of he receiver, he small difference ~ beween hese wo can easily be separaed wih a low-pass filer from he much larger differences exising beween fa and he oher muliples of fe and which are abou equal o fe or several imes fe (hus 4 Mcfs or higher). f he frequency fa of he oscillaor under es changes by an amoun Lfa, hen ~ changes by he same amoun" and his amoun is usually considerable compared wih ~. How his drif is deeced and recorded will be shown below. Le us now consider he case where fa lies in he neighbourhood of 24fe,' and le he difference again be ~. This difference will hen be presen already a he oupu of he firs mixing valve, so ha in his paricular case here is no need of he second mixing valve. Thus he second mixing valve could in his case be by-passed, bu i has been found ha he swiches which would hen be required can be dispensed wih and he circui lef unchanged, because in spie of he valve being unblocked and blo~ked wih he high frequency fe i sill ransmis he low frequency ~ wih sufficien ampliude. By means of he second mixing process described - he mixing of j24 fe - fol wihfe - i has hus been found possible o o;anspose he drif o a much lower frequency range, wih only one fixed muliplicaion facor (24) of he crysal frequency, for en differen frequencies. Furher, i is necessary o record he relaively low oupu frequency ~ of he second mixing valve varying according o he amoun of he drif. This can be done, for insance, by generaing a direc curren which is (approximaely) proporional o he frequency ~ and causing his curren o flow hrough a recording meer. To produce such a direc curren we have made use ofaknownfrequency deecor circui. This is preceded by an amplifier (A 2 in fig. 2) and a limier. The amplifier is needed because he oupu volage of he second mixing valve is oo small for he frequency deecor o funcion properly; moreover i conains he filers blocking he volage componens wih undesired high frequencies. The limier clips he peaks of he amplifier oupu, so ha an almos square wave volage of consan ampliude is produced. The laer is necessary because he oupu' curren of he frequency deecor has o be independen of he ampliude of he oupu volage from he second mixing valve. From he fac ha he meer records.only he absolue value of he difference beween one of he sandard frequencies nfe and he oscillaor frequency fa i follows ha he recorded curve (e.g. fig. 3a) leaves one in doub as o wheher fa was smaller han ufe (fig. 3b) or greaer (fig. 3c). This quesion can be solved by slighly urning he uning knob of he receiver such as o make fa somewha larger and hen noing in which direcion he deflecion of he meer is hereby changed; if he meer reading is reduced hen fa was smaller, han ufe; if i is increased fa was greaer.

4 196 PHLPS TECHNCAL REVEW VOL. 12, No. 7 fó= )n{:-{,j M=e+Óo g &01 óol e g, ++ -, nfi:-~. nfc- h. óol ~ nfi:-~ - nh s & Ó o J Fig. 3. a) Example of a recording. The frequency drif L.1f a he momen = ' is he reading a ' less he original deflecion 6 0, The variaion of nf, - fu may have been according o (b) or (c); i is easily deermined which of hese wo possibiliies i was by purposely increasing fa a lile. A urning poin in he recording (d) is o be ascribed o nfe-fo passing hrough zero according o (e) or according o (f). n (d) he drif a he momen = ' is obained by adding he iniial value (jo o he deflecion E. The recorded curve mayalso show a urning poin on he zero line (fig; 3d), his being he case when he difference beween nfe and fo has passed hrough zero during he es. The rend of nfe-fo may hen have been according o fig. 3e or fig. 3f. By purposely increasing fo a lile i can be decided which of he wo applies. The manner in which he ampliude of he frequency drif has o be read from he recorded curve is indicaed in figs 3a and 3d. : quency fe does no change by more han 1 : 10 6 per C emperaure variaion. This is so lile ha in our case i is no necessary o place he crysal in a hermosa, and his is forunae in ha i avoids he long waiing ime involved in hermally balancing he hermosa. By providing ample venilaion openings in he casing conaining he apparaus and also proecing he crysal wih hea-insulaing maerial, i has been possible o limi he final value of he frequency drif of he crysal o abou 40 cs and hus he drif of he sandard frequency o abou 1000 cs, which is sufficienly small for our purpose. The funcioning of he crysal oscillaor and he muliplying sages can be checked by measuring he grid curren of he valves employed in hese sages, for which purpose a micro-ammeer wih swich is provided. Amplifiers and mixing sage n fig. 4 a somewha simplified diagram is given' represening he circui of he firs amplifier (A in fig. 2, wih an EF 40 valve) and he second mixing valve M 2 (he hexode par of an ECH 41 valve). Beween hese wo valves is a circui, which can be uned o he frequencies fe, 2fe, 3fe 4fe and Sfe, according o wheher he drif is o be measured a fo ~ 23fe or 2Sfe, 22fe or 26fe,." 20fe or 28fe, or 19fe respecively. n hese five cases hè absolue bandwidh has o remain he same, his being achieved by swiching only he inducance and leaving he capaciance and parallel resisances unchanged. The circui is so arranged ha he gain a; a frequency deviaing 0.2 Mcls from.he resonance frequency (he drif can be measured o 0.1 Mcls maximum) is no more han 3 db less han a resonance. For measuring a Form of execuion We shall now briefly discuss some deails of he various componen pars. Crysal oscillaor Owing o use being made of a quarz crysal wih a "emperauré-independen" cu, he fre Fig. 4. Amplifier Al of fig. 2, wih he mixing valve, M 1 of he receiver and he mixing. valve M 2 belonging o he drif meer. Beween he penode EF 40 and he hexode ECH 41 is a uned circui formed by he capacior C and one of he coils L 1 L5 according o he desired value of n; for n = 24 he inducance branch is inerruped. '

5 JANUARY 1951 RECORDNG FREQUENCY-DRFT-METER 197 fo F:::: 24fc he inducance branch is inerruped, so ha an ordinary resisance-capaciance coupling is lef. Limier and frequèncy deecor The circui of he limier and of he frequency deecor is represened in fig. 5. The limier works wih wo diodes (Dl' D 2 ). The cahode of Dl is kep a a consan poenial E F::::+ 20 V. A L FD in series wih D4 are a resisor Ra and a micro-ammeer, hese being shuned by a smoohing capacior C 5 n order o avoid he possibiliy of he diodes causing a cerain zero curren o flow..hrough he meer while no exernal volage i~' 'applied, he cahode of Da is biased wih a small,posiive poenial Eo F:::: 2 V. The formula for he curren flowing hrough he meer is: 1- (JC a (E-E).. () -+(JCR 0' a a where (J is he frequency of he pulsaing volage a he erminal b, Ca he value of he' capacior conneced in series wih he diodes, while Ra includes he resisance of he micro-ammeer. The formula (1) can be derived in he following way, A /'. ignoring he volage loss in li'e diodes.. f When he volage a he erminal b is zero, in he saionary ---~ ----""'---+--H ~- ' sae here is a volage across Caequa1 o he volage V =. +25OV +20V +KJV +2V Fig. 5. Circui diagram of he limier (L) and of he frequency deecor (FD). EL 41 oupu valve of he amplifier A 2 (fig. 2), wih he anode resisor Ra. Dl'" D, diodes EA 50. Cl' Ca coupling capaciors, Ca' C 4, Cij smoohing capaciors, J.',A micro-ammeer wih resisor Ra. The funcion of he resisor Ra is o keep he mean poenial of he poin b a + 10 V. A he poin a here is pracically he same alernaing volage as a he anode of he oupu valve in he amplifier A 2 ; he ampliude of his alernaing volage is much greaer han E. The diode Dl passes curren only when he alernaing volage a a is greaer han E, and he diode D 2 conducs when he poenialof a is negaive. Disregarding volage loss in he diodes, he poenialof he poin b is hen + E and zero respecively. The ransiion beween hese wo values akes place in he very shor inervals during which no curren flows hrough eiher of he diodes. Thus he limier supplies an almos square-wave oupu volage, wih alernaely he values +E and zero. Since b is conneced via he resisor R 2 o a poin wih he poenial ie, he posiive peaks of he alernaing volage a a are cu off as much as he negaive peaks, so ha he inervals during which he volage a' b assumes he values +E and zero have he same lengh. The frequency deecor, which is of a known ype 2), similarly has wo di~des (Da,D 4 ). Conneced a) S. W. Seeley, Ch. N. Kimbal and A. A. Barco, Generaion and deecion of frequency-modulaed waves, R.C.A. Review 6, , 1942; reprined in Frequency Modulaion, Vol., pp. 147 e seq. (R.C.A. Princeon, N.. J., U.S.A.). Ra across Cij' When he volage a b suddenly changes from zero o +E hen Careceives via he diode Daa charge equal o Ca(E-V-Eo)' When he volage a b drops again from +E o zero his charge flows via he diode D, o Ra and Cijconneced in parallel. This akes place j imes per second, so ha he mean curren amouns o = j Ca (E - Ra - Eo). Solving his equaion for leads o formula (1). As can he seen from formula (), he direc curren is pracically proporional o he frequency (Jif care is aken ha (J CaRa < 1. This is wha is in fac done in cases where he deecion has o be as linear as possible. n our case, however, he ~eer need no record he frequency on a purely linear scale; in fac his is even undesirable, because wih high values of (J, which occur while uning fo near o one of he reference frequencies, he meer would he overloaded. For his reason Ra and Ca % _ ó Fig. 6. The direc curren flowing hrough he micro-ammeer (in % of he curren required for full deflecion) as a funcion of he frequency j in kcls a he inpu of he frequency deecor..

6 198 PHLPS TECHNCAL REVEW VOL. J.2, No. 7 Fig. 7. Frequency-drif meer, in a form suiable for mouning in a rack. The lef-hand meer, calibraed in kc/s, is used for preparaory adjusmens; he recording meer is no shown in he illusraion. The righ-hand meer is used o check he various grid currens seleced by he knob in beween hem. Underneah is he knob for selecing he desired reference frequency (n = ), cf. fig kcls,... ~ 80 ~ 4 4 kc/s LD Cl C'Y) ~.- ;.,~ p' ---, n ~ ,-.." -\. 10.'. ~" e..a '-., Fig. 8. Two recordings. The firs shows a jus accepable drif of = 43 kc/so The second recording shows a urning poin five minues afer he sar, and a he end of 15 minues a maximum a which he oal drif is = 85 kc/so Afer a second urning poin a second maximum is reached, a abou hree hours from he sar; i shows a frequency drif of = 40 kc/s wih respec o he firs maximum. f only he difference beween- he frequencies a he sar and ha a a momen hree hours laer had been measured he accepable amoun of 55-} 0=45 kc/s would have been found, bu he quie unaccepable maximum drif of 85 kc/s would have remained unnoiced... 1h

7 :JANUARY 1951 RECOR?NG FREQUENCY-DRFT-METER 199.have been so chosen ha wih {J = 100 kc/s he value of {J C 3 R 3 = 1. The curve = f( (J) hen assuines he shape represened in fig. 6. There hus is no longer any risk of overloading, while he frequency is sill sufficienly legible over he whole range, especially so wih small values of {J, so ha even a small frequency drif can sill be. accuraely deermined. n he form in which i is shown infig. 7 he drif meer has w o micro-ammeers conneced in series wih he resisor R3 (fig. 5): a normal buil-in meer and a recording insrumen (full deflecion a 34 [LA) kep separae. While he es is being sared he readings are aken from he buil-in meer. The frequency scale of he buil-in meer runs up o 150 kc/s, and ha of he recording meer o 100 kc/so Finally, in fig. 8 wo recordings are reproduced, one showing an accepable drif and he oher an inolerably large drif. Summary. The frequency-drif meer described here has been designed for measuring he frequency variaion of he oscillaor in various ypes of superheerodyne F.M. receivers in he Mc/s band. The drif meer is conrolled by a quarz crysal wih frequency fe = Mc/s. of which he 24-fold frequency (101.6 Mc/s) is obained wih he aid of muliplying sages. n he mixing valve of he receiver his harmonic is mixed in he usual way wih he oscillaor volage having he frequency fo ha is o be observed. The oupu volage, which conains. iner alia. he bea frequency 124fe - fol. is mixed wih a volage of he crysal frequency in a second mixing valve incorporaed in he drif meer and so adjused ha i is rendered conducive only a he peaks of he laer volage. The resul is ha from his second mixing process frequencies such asl24fe-fo ± kfel = 1(24 ± k)fe-fol. wihk= are obained. so ha. wih only one muliplier working wih good seleciviy, a series of reference frequencies. (24 ± k) fe. are available which can be separaed by means of a circui uned o one of he frequencies kfe. n his way. in he drif meer described en reference frequencies are available. from 19fe = 80.4 Mc/s up o and including 28fe = ll8.5 Mc/s. Afer amplificaion and limiaion. he oupu volage of he second mixing valve. wih frequency 1(24 ±' k)fe - fol. produces a he oupu of a normal frequency deecor a direc curren which is approximaely proporional o he frequency variaion and hus also o he drif of fo. This direc curren is recorded as a funcion of ime by a recording meer>and hus a grea deal of ime in he regular checking of he drif of receivers is saved.