MICROPHONY IN ELECTRON TUBES

Transcription

1 1960/61, No MICROPHONY IN ELECTRON TUBES y S. S. DAGPUNAR *), E. G. MEERBURG **) nd A. STECKER ***) : Microphony my e defined s the occurrence of n electricl interfering signl produced s result of mechniclor cousticl virtions of circuit element, e.g. n mplifying tue. The effect is s old s the rdio tue itself. At first it could e kept within Olmds y mounting the tues in resilient holders. At the levels of mplifiction common nowdys, however, this simple mesure isfr from sufficient. Theoreticl nd experimentl investigtions hve shown. wht cn e done in the design nd construction of tue to minimize microplumie effects. The rticle elolv,which emodies contriutions from British, Dutch nd Germn lortories, gives some ide of these investigtions nd of the progress mde in recent yers in comtting microphony, Introduetion Amongst the component prts of rdio sets, mplifiers, etc., there re mny tht do not constitute mechniclly rigid ssemly, ut consist of prts cple of physicl virtion t frequency generlly within the udio region. As the prts virte the distnce etween them lters, nd this is ccompnied y fluctutions in the electricl properties of the circuit element involved. Tke, for exmple, vrile cpcitor: if the pltes virte with respect to one nother, the result is periodic vrition in the cpcitnce. If the cpcitor is prt of the tuned circuit of n oscilltor, the frequency of the generted voltge will lso vry periodiclly, i.e. it will he sujected to frequency modultion, giving rise to interference in the output signl. This production of n interfering signl s result of mechniclly virting components is known s microphony. Electron tues re prticulrly suject to microphony, nd in this rticle we shll e concerned solely with mierophonic effects in electron tues. Physicl virtion of the electrode ssemly not only cuses vritions in the epeitnees etween the electrodes ut lso fluctutions of the node current nd mutul inductnce, nd hence directly ffects the gin of the tue. There re mny cuses of virtion in n electron tue. Aprt from incidentl virtions or shocks, there re those to which cr rdios, trnsceivers, rdio equipment in ircrft, etc., re constntly sujected, there re the virtions due to the motor in grmophones nd tpe recorders, the mechnicl shocks cused y the epertion of switches in vrious equipment, nd ove ll the virtions *) Mullrd Rdio Vlve Co., Ltd., Mitchm, Englnd. **) Electron Tue Division, Philips, Eindhoven. ***) Development Lortory of Vlvo GmH, Rdioröhrenfrik, Hmurg. cused y the loudspeker. Loudspekers re often plced very close to mplifying tues nd cn trnsmit virtions to the ltter oth cousticlly (vi the ir) nd mechniclly (through the cinet, the chssis nd the tue holders). This sitution is prticulrly dngerous in tht the loudspeker itself reproduces the interfering microphony signl; if the gin is sufficiently high, this my give rise to cous-. tic feedck ("howling"), nd if not, it my in ny cseproduce troulesome reverertion. Microphony cn lso' produce severe interference in television receivers. Loudspeker virtions here my e trnsmitted to mplifying tues in the high frequency, intermedite frequency or video frequency prt ofthe receiver, cusing troulesome fluctutions in the rightness of the picture. Microphony in tues in the deflection circuits my distort the picture s well s cuse displcements in lines. lil recent yers extensive investigtions hve een crried out in mny lortories oth into the requirements to e met y tues in modern equipment in order to minimize microphonic effects, nd into the mesures tht cn e dopted to mke the tues fulfil these requirements. This rticle will del with the work done long these lines in vrious Philips lortories nd the results otined 1). 1) See lso the following pulictions: B. G. Dmmers, On the microphony of the EF 86, Electronie Appl. 16, , 1955/56; B. G. Dmmers, A. G. W. Uitjens, E. G. Meerurg nd M. A. de Pijper, Reflections on microphony, Electronic Appl. 18, 15-18, 1957/58; B. G. Dmmers, A. G. W. Uit jens, K. Hoefngel, E. G. Meerurg nd M. A. de Pijper, -Cuses nd effects of microphony in the R.F. nd I.F. stges of television receivers, Electronic Appl. 18, ,1957/58; A. Stecker, Die Mikrofonie der Elektronenröhre - Theorie und Anlyse, Vlvo Berichte 4, 1-21,1958 (lso Electronic Appl.18, , 1957/58); H. Hellmnn, Die. Prüffeldmessung der Mikrofonie von Elektronenröhren, Vlvo Berichte 4, 22-35, 1958; D. Hoogmoed, Microphonic effects in electron tues, Electronic Appl. 19, 25-44, 1958/59.

2 72 PHILIPS TECHNICAL REVIEW VOLUME 22 Fctol's determining the strength' of the microphony An electron tue sujected to cousticl nd/or mechnicl virtions undergoes periodiclly lternting ccelertion. It is the mgnitude of this ccelertion tht primrily determines the strength of the microphony. To give n ide of the ccelertions involved, it my e mentioned tht mesurements with virtion pick-ups in rdio nd television receivers hve shown 2) tht loudspeker fed with power of 50 mw gives rise to tue ccelertions from O.lg to 0.25g (g = ccelertion of the force of grvity). A higher power evidently cuses greter ccelertions, the increse eing proportionl to the root of the power. In cr rdios the ccelertions produced y engine virtions re much greter thn those cused y the loudspeker. Of course, the type of cr, the stte of the engine nd other conditions re importnt in this respect. Tests mde on the instrument pnels of numerous types of crs hve shown tht, under certin circumstnces, ccelertions up to 25g my occur. Aprt from the mgnitude of the virtions to which the tue s whole is sujected, the ex~ent to. which the virtions re trnsmitted from the se or wll of the tue to the electrodes lso hs n importnt ering on the strength ofthe microphony. Further fctors involved re the stiffness of the components nd the rigidity of their mountings. A further point to e tken into ccount in this connection is the function of the tue in the pprtus concerned, since this function determines the prmeter whose fluctutions my prove most troulesome. For instnce, where tue is to e used in low-frequency mplifier, chnges in the cpcitnces etween the electrodes will seldom e importnt, wheres vritions in the node current s result of electrode virtions my he very importnt indeed, since these vritions, fter mplifiction, re usully pplied to the loudspeker nd mde udile. Cpcitnce vritions, on the other hnd, cn e very troulesome in the oscilltor tue in superheterodyne receiver, prticulrly if the receiver is tuned to high frequency. In tht cse the circuit cpcitnce is smll, nd s result the tue cpcitnces hve considerle effect on the frequency of the voltge generted y the oscilltor. Periodic vrition of these cpcitnces thus gives rise to frequency modultion which, in n FM receiver, is herd through the loudspeker. This my lso e the cse in n AM receiver if the set is not exctly tuned to the received signl. Vritions in the frequency of n IF signl then give rise to 2) See the rticle y Hellmnn under 1). mplitude modultion which, fter detection, gin results in n interfering low-frequency voltge. Amplitude 'nd frequency modultion mylso e cused y cpcitnce vritions in one or more of the rdio frequency or intermedite frequency circuits of receiver, giving rise to fluctutions in the mgnitude nd phse of the output voltges of the mplifier stges involved. In cses where microphony cuses fluctutions of mutul conductnce, the effect cn he troulesome if the tue is used in the rdio freqûency or intermedite frequency circuits of n AM receiver, since periodiclly vrying mutul conductnce results in vrile gin, nd thus modultes the RF or IF signl voltge in mplitude. Microphony in tue is more troulesome' the more mplifier stges re connected ehind the tue, in which cse correspondingly smller vrition in one of the prmeters of the tue will e sufficient to produce n impermissily lrge lternting current to the loudspeker. Inconsistent nture of microphony Becuse of the numerous fctors governing its strength, microphony in prctice is n irregulr, inconsistent phenomenon. A tue fulfilling certin function in prticulr pprtus my give no difficulties, wheres in nother function or nother pprtus it my exhiit excessive microphony. The loction of the tue nd the position in which it is mounted mylso hve considerle influence. Moreover, individul tues of the sme type my show mrked disprities. In spite of the extremely nrrow tolernces used in the mnufcture of cornponents, it is impossile to void slight construetionl differences from tue to tue. This hs no significnt effect on the purely electricl properties of the tue, ut it my give rise to considerle differences s fr s microphony is concerned. Consequently, certin prcticl methods of testing cn only e crried out on sttisticl sis; whether prticu1r modifiction introduced in tue will improve the tue's microphonic ehviour in prc-.rice cn only he estlished y investigting firly lrge numer of individul tues. The inconsistent nture of microphony is ccentuted y the fct tht the frequency spectrum ofthe virtions to which the tues re sujected in prctice is extremely irregulr in shpe. The reson for this is tht the chssis, the cinet nd other structurl elements of electricl pprtus exhiit mny different resonnce frequencies for mechnicl nd cousticl virtions, so tht the whole ssemly ehves s if it consisted of lrge numers of

3 1960/61, No. 3 MICROPHONY IN ELECTRON TUBES 73 Fig. 1. Frequency spectrum of the ccelertion to which tue in certin type of rdio receiver is sujected when the loudspeker is driven y constnt power of 50 mw t vrying frequency. mutully coupled resontors. Fig. 1 shows n exmple of frequency spectrum of the ccelertion undergone y tue in rdio receiver when constnt electricl power of 50 mw is supplied to the loudspeker t vrile frequency. A frequency spectrum of this kind is otined y sustituting for the tue virtion pick-up, mounted in continer whose dimensions nd weight correspond pproximtely to those of the tue. Where three pick-ups re used, mounted in directions perpendiculr to ech other, one cn lso determine the direction in which the ccelertions occur. Such comintion of three virtion pickups is shown in fig. 2. The whole ssemly is roughly s hevy s n electron tue nd cn e inserted 111 one of the tue holders in the pprtus under test. Methods of investigting microphony There re vrious direct methods of investigting microphony tht cn esily e crried out without specil equipment. For instnce, the microphonic tendency of n udio mplifying tue cn e scertined y incorporting the tue in n mplifier circuit. The output voltge is pplied vi L I I I J A 2371 Fig. 3. Principle of simple set-up for investigting mierophony in n udio-frequency mplifying tue. B tue under test, L loudspeker, A vrile mplifier. Fig. 2. Comintion of three virtion pick-ups, used for mesuring the virtions to which tues re sujected in electronic equipment. The whole ssemly cn e fitted in tue holder in plce of tue vrile mplifier to loudspeker set up ner the tue. This rrngement is shown schemticlly infig. 3, where B is the tue under test, L the loudspeker nd A the vrile mplifier. The gin of A is cljusted until it is just sufficient to cuse coustic

4 PIIILIPS TECHNICAL REVIEW VOLUME 22 oscilltion, fter which the "sensitivity" of the comintion of B nd A t this setting is found. This is generlly tken to men the lternting voltge required on the control grid of B in order to produce n output of 50 mw from A. The gin setting of A so found is clerly too high when us cd with' the tue B; it is thus possile to give sensitivity of the comintion tht is permissile, to guide users of this type of tue. Oviously, specifiction of this kind is useful only in circuit rrngement exctly corresponding to tht with which the experiments were crried out. A smll constructionl chnge in the pprtus in which the' tue is used cn considerly lter the tendency to microphony. For this reson, nd ecuse of the ove-mentioned spred etween individul tues of the sme typc, it is lwys necessry to llow wide sfety mrgin. A rdio-frequency tue cn he tested in similr wy. The tue is incorported in, n RF mplifier stge nd n unmodulted RF signl voltge is pplied to the control grid (see fig. 4). A detector is _j I I 2372 Fig. 4. Principle of set-up for investigting microphony in rdio-frequency mplifying tue. B tue under test, L loudspeker, A vrile mplifier, D detector, HF signl genertor. connected to the output of this mplifier stge, whichis gin followedy vrile udio-frequency mplifier nd loudspeker. Microphony now cuses modultion of the RF voltge, nd the detector delivers n udio signl which, vi the AF mplifier nd the loudspeker, cn produce coustic oscilltion. The methods descried cn e used for compring different types of tues or individul tues of the sme type, nd lso for checking the results of modifictions mde in tue to reduce microphony. They give no indiction, however, s to which cornponents in tue cuse the microphony, nd re therefore no help s regrds the introduetion of the necessry improvements. In this respect nother method is helpful. Insted of mking the set-up howl, the loudspeker is connected to seprte signl genertor nd mplifier, nd the output voltge of the tue is mesured with the id of n mplifier nd vcuum-tue voltmeter (fig. 5). This 'mkes it possile to choose Fig. 5. Principle of set-up for investigting microphony in electron tues. The loudspeker L is fed vi the mplifier A 1 with voltge from the signl genertor TG. The signl voltge due to microphony in tue B is pplied to vcuum-tue voltmeter BV vi the mplifier A 2 nd to vry the frequency of the virtions to which the tue is sujected. The strength of the microphony is then found to vry quite irregulrly with the frequency. This is prtly due to the fct tht the components 'of the electrode system hve different resonnce frequencies for mechnicl virtions, so tht the tue ehves s if it consisted of lrge numer of mutully eoupled resontors. A further fctor, however, s mentioned ove nd illustrted in fig. 1, is tht, even where the loudspeker power is constnt, the ccelertion undergone y the tue shows highly irregulr spectrum. As result, it is not esy to study the microphonic properties of electron tues with set-up s in fig. 5: there is lwys the possiility tht the cuse of strong microphony occurring t prticulr frequency my lie outside the tue itself. To rrive t results tht re governed solely y the tue we must therefore set the tue in virtion directly nd not vi loudspeker, cinet nd chssis. One method of chieving this is to suject the tue to n impct of given strength nd to mesure the resultnt microphonic signl voltge. An pprtus designed for this purpose is shown in fig. 6. Even here, however, the results re not very stisfctory. A low rings ll ëomponents of the tue simultneously into virtion, nd only the totl result cn e mesured from the signlvoltge therey generted. Consequently, this method too is relly only suitle for compring tues one with the other, nd not for trcing the cuses of microphony. A thorough study of the microphonic properties of 'tues demnds tht the tues e sujected to virtions of constnt ccelertion nd vrilefrequency. Only then is it possile to drw con-

5 1960/61, No. 3 MICROPHONY IN ELECTRON TUBES 75 Fig. 6. Apprtus for studying the microphonic properties of electron tues y sujecting the tue to known impct. is proportionl to the deflection of the virtor. These voltge vritions cn e mplified nd mesured. It follows from the theory of hrmonic virtions tht, if the pek ccelertion is to remin constnt, the mximum deflection must e inversely proportionl to the squre of the frequency, i.e. with incresing frequency it must decrese y fctor of 4 per octve. With the virtor descried here this ws indeed found to e the cse in the required frequency rnge. The wy in which the tue under test is fixed to the virtor clls for prticulr cre. Strictly speking, it should e perfectly rigid, otherwise the virtions of the coil former will not e trnsmitted to the tue completely independent ofthe frequency. Now, some resilience in the mounting of the tue is unvoidle nd consequently the tue resontes t the frequency of the mounting. To prevent clusions, or t lest inferences, s to the cuse of strong microphony t prticulr frequency. With this oject in view, the constnt ccelertion hs een chieved y testing the tues in specilly designed virtor. This method hs for some time now een pplied in severl Philips lortories to numerous types of tues, nd will now e discussed in somc detil. A vihrtor for the study of microphony Fig. 7 shows n xil cross-section through virtor designed for investigting microphonic effects in electron tues. The construction of the virtor closely resemles tht of n electrodynmic loudspeker. It consists primrily of coil which cn move in the ir gp of ring-shped mgnet. The coil is wound on n luminium former, which is supported in sleeve erings. The resonnce frequency of the whole ssemly is in the region of 30 kcjs, which is gret del higher thn the highest frequency of the rnge in which microphony tests re usully mde (30 to cis). In this frequency rnge, then, where the lternting current in the coil is constnt n lternting ccelertion of lmost constnt pek vlue is otined. (An lternting current of 100 ma ws needed for pek ccelertion of 19.) This cn e quite esily checked y mounting sttionry metl plte short distnce from the upper surfce of the coil nd y mesuring the vritions occurring in the cpcitnce etween this plte nd the coil, this cpcity eing inversely proportionl to the distnce. For constnt chrge, voltge vritions pper cross the cpcitor thus formed, the mgnitude of which Fig. 7. Cross-section of virtor for studying microphonic effects in electron tues. S coil, SL coil former, L sleeve erings, M mgnet, PI nd P 2 pole pieces, TO virtion pick-up, HS dptors for clmping the tue to the coil former.

6 76 PHILlPS TECHNICAL REVIEW VOLUME 22 Fig. 8. Upper prt of virtor fitted with tue to e sujected to virtions: ) long the xis; ) perpendiculr to the xis. this ffecting the mesurements, the virtor nd mountings must e so designed tht the resonnce frequency is well ove the frequency rnge under investigtion. Some components used for this purpose re shown in the upper prt of fig. 7. To enle constnt check to e kept on the virtions, virtion pick-up of the piezo-electric type is fitted (with piezo element of rium titnte); this is denoted y TO in fig. 7. This device lso indictes whether insufficiently rigid mounting of the tue is cusing spurious resonnce. Fig. 8 shows two photogrphs of the upper prt of the virtor, with tue mounted in two positions, enling it to e sujected to virtions prllel or perpendiculr to the xis. Mesurements with the virtor With virtor s descried ove tue cn e sujected to known ccelertion which, s opposed to the methods illustrted in figs. 3,4 nd 5, is independent of the incidentl resonnce frequencies of other prts of the pprtus. If strong microphony ppers t prtienlr frequency, it is now cler tht this frequency corresponds to the resonnce frequency of one of the tue components. Detection of these frequencies is ccomplished y connecting the tue y flexile wires to n mplifying circuit: the signl voltge produced in this circuit s result of microphony is mesured whilst the virtor frequency is slowly vried. The mesurements re fcilitted y using recorder. Exmples of spectrogrms otined in this wy re shown in figs In these nd similr mesurements the virtion frequency should e vried slowly nd very evenly, the mechnicl virtions of the vrious components eing very little dmped. Becuse of the wek dmping, the vrious resonnces occur only very ner to the exct resonnce frequency: mny peks in the spectrogrm re so shrp tht they might esily e missed. Once it hs een found tht strong microphonic effect occurs t prticulr frequency, the next thing to do is to trce the component responsile for it. There re vrious wys of setting out this. One ovious method is to clculte the resonnce frequencies of components whose very slight movements cn e expected, on theoreticl grounds, to hve considerle effect on the electricl chrcteristics of the tue. One cn then scertin whether one of these frequencies coincides with pek in the spectrogrm. If this is so, it is resonle to ssume tht the component in question must e the cuse of this pek. Further experiments re then needed to show whether this ssumption is correct or not. In prctice, this method turns out to e most unstisfctory. The min reson is tht clcultions of the resonnce frequencies of components in n electrode system cn seldom e more thn rough pproximtions. Exct formule cn e derived only for simple configurtions, nd the ppliction of such formule to prcticl cses clls for pproximtions nd corrections; lso, it is generlly not ccurtely known just how the vrious electrodes re clmped or supported, or whether there is ny ply etween them. We shll illustrte the ove method nd its shortcomings y tking grid s n exmple. The conventionl grid construction is shown in fig. 9. Two uprights (or "hckones") 51 nd 52 re mounted in holes in the mic discs M 1 nd M 2 The grid wires D re wound heliclly ronnd the uprights. If we now regrd 51 nd 52 s freely vihrting rods, their resonnce

7 1960/61, No. 3 MICROPHONY IN ELECTRON TUBÉS 77 These formule hold good only when nd R re roughly the sme (RI < 2). Consequently, nd lso ecuse the ctul shpe of the grid wires never exctly stisfies fig. lo or, the result here too cn never e more thn very roug~ pproximtion. Another source of uncertinty is the fct tht mny elements cple of mechnicl virtion in n electrode system re coupled to one. nother, resulting in resonnce frequencies tht do not correspond to those of the elements individully. Strooscopic exmintion 2375 Fig. 9. Simplified construction of grid. SI nd S2 uprights ("ckones"),d grid wires,mi n:llw 2 mic strips onwhichthe ssemly is mounted t the points,, c nd d, A led-in wire. frequency Ir for mechnicl virtions cn e clculted from the formul: d Ir = 412 VE ek. Here d is the dimeter nd I the length of the rod; E is the modulus of elsticity nd (] the density of the mteril, nd K is constnt which depends on the wy in which the upright is held. The mgnitude of this constnt is: K = 0.56 if the virting rod is clmped t one end nd free t the other, K = 3.56 if the rod is clmped t oth ends, K = 2.45 if the rod is clmped t one end nd held such tht it cn pivot t the other, K = 1.56 if the rod is held such tht it cn pivot t oth ends. Owing to the unvoidle spred in the dimensions of the grid uprights nd of the holes in the mic supports, it is never certin whether the uprights t positions,, c nd d shoulde regrded s clmped, pivoted or free. Extremely smll differences in dimensions, which my hve no perceptile effect on the electricl properties of the tue, my hve mrked effect,in view ofthe differencesin K,on the resonnce frequency of the grid uprights. A further inccurcy in the clcultion is due to the presence of the grid wires D. Their effect cn e llowed for s n increse in the mss of the grid uprights, ut this is oviously rough pproximtion. Finlly, the fct tht connection wire A is ttched to one of the uprights cn lso only e tken into ccount y very rough pproximtion. For the grid wires two empiricl formule hve een worked out 3) which pply to wires ent in the form of n re (jig. lo) nd in the form of rectngle (fig. lo). The formul for grid wire s in fig. lo is nd for wire s in fig. lo: d 1IE Ir = R2 le' d lie Ir = R2 V e. 3) See P. M. Hndley nd P. Welch, Vlve noise produced y electrode movement, Proc. Inst. Rdio Engrs. 42, , The only wy to point with certinty t one of the components of the tue s the source of strong microphony t prticulr frequency is to oserve directly tht this component in fct resontes t tht frequency, i.e. virtes with lrge mplitude. Since this "lrge" mplitude is not usully perceptile to the nked eye, it must e oserved under microscope. In order to mke it possile to oserve grids etc., it my further e necessry to mke numer of tues with specil openings in the node or in the screening. As rule the frequencies t which the investigtions re crried out re too high for the eye to e Q 12 _. R --0 s 0 -G.e Fig. 10. ) Grid wire ent in the form of n ure, ) rectngulrly ent grid wire. le to follow the movement directly. The movements cn e mde visile, however, y Illuminting the tue with strooscope. The virtor nd the strooscope re then fed y two signl genertors delivering lternting voltges whose frequencies differ y few cjs. An rrngement designed for the purpose is shown schemticlly in

8 78 PHILlPS TECHNICAL REVIEW VOLUME 22 fig. 11. When the tue is set in virtion, the prts in question cn e seen under the microscope to virte t frequency equl to the difference etween the frequency of virtion nd the illumintion frequency. The fct tht one of the prts Fig. 11. Block digrm of n rrngement for strooscopic investigtion of microphony in electron tues. TG l nd TG 2 signl genertors, Al nd A 2 mplifiers, T virtor, B tue under test, SL strooscope lmp, Mi microscope. is resonting t the pplied frequency is mnifested not only y the increse in mplitude ut lso y the phse reltionship etween the impressed force nd the deflection. When virting system is driven y forcewhose frequency is much lower thn the resonnce frequency of the system, the deflection is in phse with the force. If the frequency of the force is much higher thn the resonnce frequency, the deflection nd the force re in ntiphse. The trnsition from one of these sttes to the other tkes plce in frequency rnge round the resonnce frequency. Theless the virtions ofthe system re dmped, the nrrower is this frequency rnge. At the resonnce frequency itself the phse shift etween force nd deflection is 90.' To produce utomticlly difference etween the virtor nd the strooscope of few cis t ll frequencies, one might couple the tuning mechnisms of the two signl gener tors. It is very difficult, however, to mke the coupling in such wy tht the frequency difference remins sufficiently smll over the whole frequency rnge of interest. If the difference is too gret the eye cn no longer follow the individul virtions. When one of the components is then excited into resonnce, the resonnce will e scrcely perceptile. An improvement in this respect is otined if the strooscope frequency is mde exctly equl to the virtion frequency y connecting the virtor nd strooscope to common signl genertor. Of course, the virting prts then pper to e sttionry, the movement eing frozen t ll frequencies. When the frequency is slowly vried, however, nd psses the resonnce frequency of component within the field of view of the microscope, the phse shift of 180, mentioned ove, cn e seen to tke plce in the virtions undergone y this component. As the frequency moves through very smll rnge, this prt is then seen to mke single hlf-virtion nd then stnds still gin. The concentrted ttention required to oserve this phenomenon is serious drwck, however, to the ppliction of this method in lrge-scle investigtions. High demnds re lso mde on the equipment; the frequency must e vried extremely slowly nd continuously. The strooscopic method ws not relly success until n pprtus hd een designed with which it ws possile, in the whole frequency rnge under investigtion, to mintin constnt difference of 1 or 2 els etween the virtion frequency nd the frequency of the strooscopic illumintion. Fig. 12 shows lock digrm of the equipment used for this purpose. As in fig. 11, the virtor T is driven y signl genertor TG vi n mplifier AI' The output of the signl genertor is gin used to operte the strooscope lmp, ut only fter first eing pplied to frequency shifter FS which delivers n output voltge whose frequency is constnt mount higher or lower thn the frequency of the pplied voltge. The output voltge of the frequency shifter is used to control the pulse genertor PG, which delivers short voltge pulses to the strooscope lmp SLo Fig. 12. Block digrm of equipment for strooscopic investigtion of microphony in electron tues. By mens of the frequency shifter FS constnt difference of 1 to 2 cis is mintined etween the virtion frequency nd the frequency of the strooscopic illumintion. TG signl genertor, Al' A 2 nd A3 mplifiers, Mot motor, PG pulse genertor, SL strooscope lmp, Mi microscope, T virtor, B tue under test, o oscilloscope.

9 1960/61, No. 3 MICROPHONY IN ELECTRON TUBES 79 'The frequency shifter consists of rotor nd sttor. The sttor is provided with three-phse winding. This is supplied with three-phse voltge derived from specil mplifier A2' fed with the signl-genertor voltge (frequency f). The rotting field thus produced induces n lternting voltge in the single-phse winding of the rotor. When the rotor is sttionry, the frequency of the ltter voltge is equl to f, ut when the rotor revolves t speed of tlf revolutions per second, the frequency of the e.m.f. induced in the rotor,winding is n mount tlfhigher or lower thnf, depending on the sense of rottion. Provided the rotor turns t constnt speed, constnt frequency difference is then mintined etween the pplied voltge nd the output voltge. The tue under test B feeds n mplifier A, the output voltge of which is pplied to one pir of pltes of n oscilloscope O. To the other pir of pltes voltge is pplied which is proportionl to the current driving the virtor. By oserving t the sme rime the picture under the microscope nd tht on the oscilloscope screen it is now possile to scertin with considerle certinty whether the virtions of prticulr component re responsile for the occurrence of strong microphony t prticulr frequency. At tht frequency the mplitude of the virtions undergone y the component in question shows mximum, nd t the sme time the lternting voltge produced y the microphony is seen on the oscillosope to rech mximum vlue. The oserver lso sees the ove-mentioned phse shift s the resonnce frequency is pssed. The phse shift lso of courseoccurs etween the current supplied to the virtor nd the voltge generted y microphony, nd is thus displyed on the oscilloscope s lissjous figure. Owing to the extremely wek dmping of the virtions the effects referred to occur in such very nrrow frequency rnge tht it is lmost impossile for two or more components to resonte simultneously, even when their resonnce frequencies lie very close together. Fig. 13 shows photogrph of set-up s here descried. With his left hnd the oserver vries the frequency of the signl genertor, whilst with his right hnd he directs the microscope nd the strooscope lmp on to the component to e exmined. Over the edge of the oculr he cn see the screen of the oscilloscope (right in figure). The equipment descried cn lso e used in nother wy. Insted of the mplified voltge output of signl genertor we cn pply to the virtor the mplified lternting voltge produced y microphony in the tue under test. In mny cses this will give rise to oscilltion t frequency corresponding to the resonnce frequency of one of the components. This component will then virte with lrge mplitude, nd it will not generlly e difficult to scertin y mens of the microscope nd strooscopic illumintion exctly which component this is. Of course, this method cn only trce the component tht mkes the mjor contriution to the microphony, since oscilltion occurs t the resonnce frequency of tht component. It is lso possile, however, to find the cuse of strong microphonic effects t other frequencies if we include in the feedck pth filter tht psses signls only in limited frequency nd. In tht cse, oscilltion cn occur only t frequency within tht nd, nd the component responsile for it cn e trced with the microscope.. In the method using n oscillting circuit it is even more importnt thn in the other methods descried tht the resonnce frequency ofthe virtor-plus-tue ssemly should lie ove the rnge of frequencies under investigtion. If tht is not the cse, the circuit might strt to oscillte t this resonnce frequency, nd the serch for the "guilty" component would then e fruitless. Exmples of microphony It is unfortuntely not possile in photogrph to give good impression of the picture oserved under the microscope when component virtes t its resonnce frequency,. Nevertheless, to give some ide of wht is seen some photogrphs re shown tht were otined y doule exposure t the extreme deflections of the virting component. With n rrngement s in figs. 12 nd 13 it is simple mtter to freeze the oserved picture of the virting component in ny desired phse. All tht is necessry is to stop the motor tht drives the rotor in the frequency shifter. Oviously, the picture is then sttionry too, nd the required phse of the virtion cn then e chosen y turning the rotor y hnd. Figs. 14 to 20 show vrious components of electron tues nd the picture seen under the microscope when the tue is mde to virte t the resonnce frequency of the respective component. The rrows indicte the direction of the virtions. Fig. 14 shows getter which, eing reltively lrge nd supported on one side only, hs low resonnce frequency, nmely 300 cis. It is evident tht component s lrge s this, though not prt of the ctul electrode system, must hve noticele effect on the opertion of the tue if set in virtion. Fig. 15 shows the two filment leds of the tue, which hve different resonnce frequencies, viz. 570 nd, 600 cis. The pictures oserved under the

10 80 PHILlPS TECHNICAL REVIEW VOLUME 22 Fig. 13. Exmining tue for microphony. The rrow points to the tue under investigtion. microscope when the tue is mde to virte t ech of these frequencies pper in fig. ls nd c. I t cn e seen tht, in ech cse, one of the two wires is virtully t rest whilst the other virtes. Fig. 16 shows the suppressor grid of pentode mde ccessile to oservtion y n opening in the node. Although the resonnce frequencies of the turns of wire differ only slightly from one nother, it cn clerly e seen in fig. 16 tht t the resonnce frequency of one of them (pprox cis) only tht turn enters into virtion. This illustrtes the fct tht the mechnicl virtions re very little dmped. Grid-wire virtions cn cuse impermissile microphony if they occur in the screen grid of pentode in the RF or IF sections of receiver. If the

11 1960j61, No. 3 MICROPHONY IN ELECTRON TUBES 81 Fig. 14. ) Getter of n electron tue. ) Picture seen in the microscope when the tue is mde to virte t the resonnce frequency of the getter (300 cjs). mutul conductnce in the pentode hs een reduced to low vlue y the utomtic gin control, the electron current psses through only few turns of the screen grid. A slight movement of one of these turns then hs considerle effect on the node current nd on the mutul conductnce. The effect is less pronounced if the tue opertes with higher mutul conductnce. More nuisnce is then experienced from virtions in the grid uprights, since this cuses lterl movement of the whole grid. Fig. 17 shows grid undergoing virtions of this kind in the triode portion of triode-hexode. Here, too, it ws necessry to cut n opening into the node. The resonnce frequency of this grid ws 1900 els. In fig. 18 the end of cthode cn e seen tht exhiited some ply in the upper mic support of the electrode system, nd therefore virted t very low frequency (600 cis). Cthode virtions re usully dmped more thn those of other components, owing to the influence of the filment with Fig. 15. ) Lower prt of the electrode system of vcuum tue. The circle mrks the ends of the filment leds; ) nd c) show the pictures of these leds seen under the microscope when the tue is successively mde to virte t the resonnce frequency of ech led (570 nd 600 cjs). c

12 82 PHILIPS TECHNICAL REVIEW VOLUME 22 Fig. 16. ) Suppressor grid of pentode, visile through n opening cut into the node. The circle mrks the prt seen under the microscope, (), when the pentode is mde to virte t the resonnce frequency of one of the grid wires (2100 cis). Fig. 17. ) Electrode system of triode-hexode. The circle mrks the grid of the triode portion, visile through hole cut into the node. ) Picture of the grid virting t its resonnce frequency of 1900 cis.

13 1960/61, No. 3 MICROPHONY IN ELECTRON TUBES 83 Fig. 18. ) Top view of the electrode system of tue. The circle mrks the end of the cthode, which showed some ply in its hole in the mic disc. ) Picture seen under the microscope when the tue ws mde to virte l the resonnce frequency of the cthode (600 cis). Fig. 19. ) Anode of n electron tue. The two prts were not fixed firmly enough t the positions indicted y the rrows, thus llowing free movement etween them. ) Picture seen under the microscope of the circled re when the tue ws set in virtion t frequency of 1300 cis.

14 84 PHILIPS TECHNICAL REVIEW VOLUME 22 Fig. 20. ) Frme grid of tue for very high frequencies. ) Picture under the microscope when the grid ws mde to virte t the resonnce frequency of one of' the wires ( cis). its insultion nd to the ermssrve coting of the cthode. Virtions of one of the structurl elements of n node t frequency of 1300 cis cn e seen in fig. 19. The reson for this virtion ws tht the prts of the node t the position denoted y the rrows hd not een properly fstened. The fct tht this method of investigtion cn lso e used t higher frequencies thn those m entioned ove is illustrted in fig. 20, which shows,-,, ' 40 :m1 6o }JV I " 20f I I. I 10 6_fJV r o _.. L_._ 1..~.. ). (. i J.. D ~) 1.". r (' I~. (,.L.I... '.._... J c. {\ (.) (" t I Fig. 21. Effect of getter construction on the microphony of n electron tue. The two constructions compred re shown on the left, The construction under results in considerle reduction of microphony.

15 1960/61, No. 3 MICROPHONY IN ELECTRON TUBES 85 the virtions of one of the wires of frme grid 4). The frequency ws cis. It need hrdly e sid tht this imposes very high demnds on the virtor nd on the rest of the circuit; the strooscope lmp, for exmple, hd to provide extremely short light pulses to produce sufficiently shrp picture. The reduction of microphony Once it hs een estlished tht prticulr component mkes sustntil contriution to the microphony of n electron tue, it is of course importnt to scertin whether structurl modifiction designed to reduce the microphony relly hs the desired effect. This cn est e checked y recording spectrogrm of the signl voltge due to microphony s function of the virtion frequency. We shll illustrte this with some exm ples of 4) The construction of frme grid is descried y G. Diemer, K. Rodenhuis nd ]. G. vn Wijngrden, The EC 57, disc-sel microwve triode with L cthode, Philips tech. Rev. 18, , 1956/57. improvements introduced, Figs. 21 to 24 show numer of spectrogrms recorded whilst the electrode systems of the tues under investigtion were sujected to lterl virtions t constnt pek ccelertion of t m/sec 2 The figures indicte the rms vlue of the lternting grid voltge producing the sme interfering signl s cused y the microphony. Fig. 21 illustrtes the result of modifying the support of getter. The upper recording ws mde on tue where the getter ws fixed to rcket which ws welded to the node t one point. This getter ws found to e responsile for the strong microphony tht occurred t frequency of 1300 cis. When the getter ws secured t two plces in the mic support, nd thus no longer connected to one of the electrodes, considerle improvement ws otined, s ppers from the lower spectrogrm. At frequencies elow 1850 cis the tue is now free from microphony. Fig. 22 shows the improvement chieved when single mounting lug on the node ws replced y. ~Of~ r+r+r I. -.~~.-'~OO~iJ\I- - _:fi..-;:'1 ' t ( I df, r J ; r - 40; ' ~ - I- I I 1_- ' ~ol40/jvl I.!.., Fig. 22. Effect of node fstening on the microphony of tue. The method of fstening with single lug () is inferior to tht with doule lug ().

16 86 PHILlPS TECHNICAL REVIEW VOLUME 22 (J I (,) 2385 Fig. 23. The effect of node design on microphony. The construction shown in (u) is gretly inferior to tht in (). doule lug, therey eliminting the originl ply in the mic. In the construction shown in the upper figure, oth prts of the node were cple of reltive virtion; s cn e seen in the lower figure, the virtion is much less pronounced with the new construction. The high pek t out 780 els in the upper spectrogrm is no longer to e seen in the lower recording. The improvement otined in nother cse, y modifying the construction of the node, ppers from fig. 23. An node consisting of two rectngulr sections, s in, is fr more rigid thn n node one of whose prts is flt, s in. The mrked improvement from the point of view of microphony is clerly to e seen in the spectrogrms. If the vrious components tht give rise to microphony cn e systemticlly trced nd improved, the microphony cn e lmost entirely eliminted, s illustrted in fig. 24. The upper spectrogrm reltes to tue which exhiited very troulesome microphony t vrious frequencies. The lower spectrogrm, recorded fter the necessry structurl improvements hd een mde, shows tht the tue is now virtully free from microphony. It is not lwys possile in the series production of tues to introduce ll the improvements tht would e desirle with n eye to microphony. Other considertions of quite different nture my often e involved, such s the effect of these improvements on the electricl properties of the tue, on the cost price or on the production tools. However, if the cuses of the microphony re sufficiently known - nd they cn nerly lwys e trced y the methods descried ove - compromise cn generlly e found tht stisfies these other requirements s well. Noise method of investigting microphouy For trcing the cuses of microphony the foregoing method yields good results. In certin cses, however, simpler nd less time-consuming method my e sufficient. This my e the cse when the purpose is not to investigte the cuses of

17 1960/61, No. 3 MICROPHONY IN ELECTRON TUBES 87 ", Bruel e. KJ"'" 5OE-' -----r-- - :..".1...,...,.,- j-' l ' I d' 1-40'- l00,.jv,~_.. I MY" ( (I t ( I (: (, ) <. I (', ~"~ ~~--~-r~~~~~~--~r-~-~~~~~~ i~~~;~e~t~ ~ ~~~~-n d,40 _loo.,uv.:.. _, ( - (, \ \,.' (\ (~ ", Fig. 24. Spectrogrms of microphony in tue, ) efore nd ) fter certin structurl modifictions to reduce microphony. microphony ut simply to compre one tue with nother in order, for exmple, to otin sttisticl dt on the effect of certin structurl modifictions. In such cses it is often enough to record spectrogrm. Using the method descried erlier, viz. sujecting the tues to sinusoidl virtions with vrile frequency, severl minutes will lwys e required to otin servicele spectrogrm. The frequency must not e vried too quickly ecuse, s exp lin ed ove, the mechnicl virtions of tue components re very little dmped, nd it is therefore likely tht some peks in the spectrogrm will e missed if the test is done too quickly. To conclude this ccount we shll riefly descrie method designed to produce quicker result. The virtor - nd the tue under test - is excited not y sinusoidl lternting current of vrile frequency ut y current contining components with ll frequencies t the sme time, i.e. current delivered y noise source. In tht cse ll components tht hve resonnce frequencies in the frequency rnge under investigtion will e excited into resonnce simultneously. If the tue is incorported in n mplifier circuit, microphony gives rise to signl composed of numerous lternting-voltge components. Mesurement of the rms vlue of this signl gives in itself n ide of the extent to which the tue in question is "microphonic", ut etter insight is otined if the signl components re mesured with selective voltmeter which gives reding in only nrrow frequency nd. By shifting this smll nd over the whole investigted frequency rnge we cn gin otin spectrogrm. This cn e done in such wy s to disply the spectrogrm directlyon ll oscilloscope screen. Fig. 25 shows spectrogrm produced in this wy. Reproducile grphs cn he otined with the selective voltmeter t I t n~ I (1 ~ IJ~rjlAt, 1'1I~',-Ili!.. ~. ij/i " n' m Ij~. P'11~!""""!'Ill.1..-::~ l1t UI po.j o n' IJ Q Fig. 25. Oscillogrm otined using the noise method of investigting microphony.

18 88 PHILIPS TECHNICAL REVIEW VOLUME 22 23QQ Fig. 26. Virtor used for microphony investigtions y the noise method. The tue under investigtion is here mounted oliquely on the virtor. sweeping the whole frequency rdge in out 15 sec. Using n oscilloscope tue with long-fterglow screen, the whole spectrum cn e seen s single disply. A virtor used for investigting microphony y the noise method is shown in fig. 26. The tue is mounted oliquely in order to otin generl picture of its microphonic properties, the tue therey virting simultneous Iy in the lengthwise nd lterl directions. A drwck of the noise method is tht the height of the peks in the spectrum depends on their width. This is ecuse the verticl deflection of the oscilloscope is proportionl to the vlue of the microphony signl, verged over the whole of the nrrow frequency nd pssed y the selective mplifier. Consequently, pek nrrower thn the ndwidth pssed y the mplifier will pper to e shorter thn roder ut otherwise eqully high pek. The picture on the oscilloscope is therefore not n exct represen ttion of the spectrum, nd this must e tken into ccount when evluting it. Prtly for this reson the noise method hs not proved gret success. If one is prepred to ccept this error, the results cn e otined just s quickly y the method using sinusoidl virtions. When the whole frequency rnge is rpidly scnned, sy in 15 seconds, here too the high peks in the spectrum wi 11 not e reproduced in their true reltionship. Even so, the resul tnt spectrogrm is still etter thn tht otined y the noise method. Becuse of this, nd the fct tht the equipment for the noise method of investigting microphony is much more cmpi icted, the system using sinusoidl virtions hs een given preference in our lortories. Summry. Vrious methods re descried for investigting microphonic effects in electron tues. Some direct methods requiring no specil circuit rrngement cn serve for compring one tue with nother, ut they give no informtion on the cuse of the microphony. For the ltter purpose virtor hs een designed y mens of which tue cn e sujected to virtion of constnt pek ccelertion nd vrile frequency. With the id of microscope nd strooscope the components responsile for the microphony cn then e trced y directly oserving their virtion. Some results chieved re illustrted y spectrogrms. Finlly, method using noise genertor is descried, where the spectrogrm is displyed on the screen of n oscilloscope.