No. 9, p; 253-284 - SEPTEMBER 1941 , Philip~· ·.Tec.hnicaI Review DEALING WITH:' TECHNICAL PROBLEMS .':RELATING, TO THE PRODUCTS, PROCESSES AND INVESTIGATIONS OF , ,N.V. pmiJ:ps' GLOErr..AMPENFABRIEKEN

EDITED BY THE RESEARCH LABORATORY OF N.V. PHILIPS' GLOEILAMPENFABRIEKEN, EINDHOVEN, HOLLAND

SEVERAL TECHNICAL PROBLÉMS IN THE DEVELOPMENT OF A NEW SERIES OF TRANSMITTER VALVES

,by'E. G. DORGELO. 621.396.615

• ' The.useof shorter and shorter radio waves involves a steady decrease i!! the dimensions of' the transmitter,valve;. When' it is at the same time desirable not to decrease the energy dissipation in the valve, various partsof the valve become relatively more heavily loaded. 'An indication: is given-in -this article ~f how several technical problems connected 'with tlÜs"u'resolved in the case ora new series of Philips transmitter valves. . .

Introduction '. : Under the influence ,of the rishig" standard' of the' wave lengths of a few metres, many of the ad- 'performance required in the field of television, a vantages of the connection al-e lost (for great, demand has '~arisen'in recent' years tor trans- a detailed discussion of the advantages and dis- - mitter valv~s which can work-with high efficiency advantages of .andpentodes we=refer to an on the wave lengths of 5 1'0 6 m 'used in teleVision. article published earlier 2). While at 'l9nger wave The existing types c~ula to some extent be used lengths screen grid and, suppressor grids function for such 'wave lengths, but usually worked at low as shielding cagé due to their constant potential, efficiency, and, moreover, they ofteninvolved great and thereby make special decoupling measures difficulties in the design and adjustment of the (neutrodyning) unnecessary,' at very short waves transmitter. this is no longer true. The self-induction 'of the con- In connection with this an entirely new series of n~ctions forms such a high impedance at these , transmitter valves has been developed, which is frequencies that the screen-grid alternating current , suitable not only for long waves but also for, the causes. the potentialof the screen grid to yary ap- short waves mentio~ed above. In appearance this preciably. type. of valve differs from the older types mainly A solution of this is to connect two valves in in its much smaller size. This small size was neces-' push-pull connection and supply the corresponding sary on the one hand to diminish the transit time ,electrodes together through a single line, so that effects, and on the other hand to make possible a' the alternating currents cancel each other in' this better adaptation to the transmitter by smaller line. There then: occurs no A.C: voltage along these electrode capacities and smaller self- and mutual connections. Between the common connection and inductions in the connections. each of the electrodes, however, there remain con: " The series of valves now developed 'consists of nections which are not in common. These connec-

I four triodes of about 250; 600, 1 200 and 2 500 W' rions mayalso still have too much impedance at , telegraphy output 1) respectively, three very short waves, so that recourse must then be had -of about 200, 500 and 1 000 Wand a push-pull to reducing their Impedance with the help of series pentode also of about 1000 W (se~fig. 1). 'resonance. The filament connection also must often The application of the push-pull principle was be tuned in this way (seefig. 2). justified by the consideration' that otherwise at These precautions are, however, at least with wave lengths of a few metres;not yet necessary when the 1) By "telegraphy output" we mean here the maximum two electrode systems are assembled in a single bulb output in class C adjustment. With this arrangement, the voltage is so strongly negative that anode current flowsfor less than half a, period. . 2) J. P. Heyboer, Philips teehn. Rev.; 2,257,1937.

, I 254 PHILIPS TECHNICAL REVIEW ve. 6, No. 9

Fig. 1. Photograph of a series of new transmitter triodes and pentodes which were devel- oped on the principles set forth in this article. The output in telegraphy adjustment. class C, for the series ofpentodes (lower row) is, from left to right: 1 000,1000,500,200 W, and for the series of triodes (upper row) it is 2 500, 1 200, 600, 250 W. and joined with the shortest possible connections. A particularly compact solution is obtained by surrounding the two and control grids with a single screen grid and suppressor grid (fig. 3). This c.oncept is realized in the push-pull pentode PPB 3/800 which can work entirely without extra tuning to the shortest wave lengths which it can reach (2.5 metres). . For the fundamental requirements which every transmitter valve must satisfy we may refer to an earlier article in this p eriodical "]. In addition to a b 392J5 the points touched upon here, which are connected Fig. 2. a,) Diagram of two pentodes in push-pull connection. with form and manner of construction of the mod- The parts of suppressor and screen grid connection not pos- sessed in common are tuned with the help of variahle series ern transmitter valves, the requirement of small capacities. The regulation of the filament impedance is by size also raises technical problems which will be means of Lecher systems of the length '/2 À. As oscillation discussed in the following sections. circuits in anode and control grid circuits Lecher systems are also used (length '/4 À). b) In the push-pull pentode PPB 3/800 only the grid and anode circuit are tuned. The filaments are connected with very short connections inside the valve. Furthermore there 3) H. G. Boumeester, Philips techno Rev., 2,115,1937. is only one common suppressor and screen grid. , ,. '.. ' ....

."i. . . " .. .' ...... I SEPT),i:MBER 1941 :NEW TRANSMI'ITîm VALVES 255

"

J' , .. _ •. ~', '" .,' ~ I> ~.. ' .. ." ""-,,: .\,. '-,, c) -Bûlb··.· ':, "', ..'. :0;" • ,~••. I ••: ... ": ,< •.

..... ;.... l·'-· .~. -.7 .... .' '" ti • ~ ; '<. • ,'T ';.~ " ;;......

., .' The' fact that great po~er is' dealt WIth ill a -. f', t .• ,c_g,"f .. ", , '" :,' .1"1. ~~all: el~ètr;dè •syst~~ '.:n.ec~ssit~~es making t~e ....--.,. 1- i-_" ,.. ., ... ~ .... coveririg, .t~e valy~, t.h!,!·b~lb:~ either very large • I 0,£ in 'diamet~:I:' or .9f.!;t' kine!.."?l glass which does hot

t-a • r easily .soften. ,W~e~ hárd glassïs;use~,the bulb. can .' < ' ',.~ be 'made quite narrow," so that'~short leads 'ivith '; .~.... ~,. .' . ,~. '~ligltt s~ff:'a;;d m:utu~i"U;_d~?ii~i{are ~btain~d .. The 'p . t ~"- üsé'of this .glass át "first prescntêd.difficulties because '. ~. . ., .. of, th,e dact that "the' existing .kinds' of hard glass . . . " t-a' becàme'" slightly .cO'nducting, 1 especially .at the .. high' ;-orking temperature. here- prevailing, 'so that . pàrticularly .the 'fused-in leàds were elèctrolytically " attacked, By making-use 'ofnèwly developed' glasses and a fusing-m' tèchniqûe borrow.èd -fróm the quartz l ....'. ~J92J6 , .: '. ~amps, hó~ê'v~f; it was. po~s~le ê~tir{!iy~to overcome Fig. 3. C~osssection of the electrode system of thc push-pull' .k pentode PPB '3/800 .• The shielding.between -the two halves. these .difficulties, ..as: 'will rbe described in the fol-

of the ánode prevents. electrons from the .righ~-~~ndcathode Iowing, .' " " , , " . J.~ , ~,' • • from reaching the left-hl,lnd anode, and wee versa. ., " i 1~~~~l~;~--''-r-r-r'-r'--''~r,M:~''-''ï'-;~-r-r-'-rII~ , .Fundamental differences between the modern .and .;W W~-J.-J.'-h++-1-+"::'·-I-HH4-++++-t-+-j-L-+-TI e !:t older types of~alves' " .' ,,:,,1_7~'--J.-=+":""t:'.:j'~'+-HH-=+-·'+'-H--I'p..f=""'=t~:£':~:::'#~~l1~'1=A ..: :We shall here ,fiFst mention those 'parts of a valve, ~ .to which specialattention must- bè paid, ap.a)3:tè~ .n, . ~ ::;:~v .. discuss two of them in Înoi:è detail. ; '., ': . ~" .' ;~,.' '.-1P;:::'=+·f--+::..--

b) , Another result of the high temperature was thé ,..1 difficulty of maintaining ~ sufficiently high 'va~u~ .. ~. ~~\ The high working temperature, which-in the case ',. ,during use. Ordinary' such as barium and of the anode may amount to 800-900 °C, prevents magnesium could not he used here for various the use of anything but pure or thoriated tungsten reasons. Asid~ from the great .hindrance to the' .for the cathode. Oxide cathodes would he too much heat radiation formed by the mirror deposited on overloaded due to heatingfromthe other electrodes. the bulb wall, the mirror also constitutes an un-" Considering the high emission of thoriated tungste!l desired and' badly reproducible capacity with re- compared' with pure tungsten (about 70 mAjW spect to the electrodes. At a bulb temperature of versus, 6 mAjW) the former has been chosen. With 300~3,50~Cthe vapour pr~ssure of barium and mag- a relatively low filament power ;;t good anode curren~ . nesium is also already so high that one may scarcely can now be obtained, so that these valves can deliver speak of a "vacuum". a satisfactory output with a fairly low anode ,:ZirconiiIm, whose getter properties have already v~ltage (3000:V orIower] .. Especially-on the short- been discussed in this periodical+), lacks the dis- est waves where the circuit losses become very , . . '. '1, great athigh voltage this is important (see jig. ,4). 4) J. p. Fast, Philips techno Rev, 5, 217, 1940.

...... '..

. , L__:...... c~~ ..:...:.._~ :__~~ --'.;_.:...... ,___ ~~ ~~~~j '._,_._' ,~,~ •. ~. ~ ,_ .. ~ 256 PHILIPS TECHNICAL REVIÉW Vol. 6, No. 9

advantages mentioned. It can easily be deposited Another great improvement was the introduetion on the electrodes in powder form and thereby auto- of bare metalleads through glass. By means of the matically take~ on the temperature necessary for correct technique of fusing in, in combination with satisfactory functioning. the use of suitable kinds of glass, it was found pos- In the following we shall go into somewhat more sible to fuse in leads of molybdenum and tungsten detail about the bulb and the getter. entirely free of oxide. Such bare leads are distin- guished from the earlier leads by the fact that even The bulb and the electrode leads through the glass. at working temperatures of 400 oe and more they remain absolutely vacuum tight. As already mentioned, electrical conduction , Leads through glass of this type are used in the .easily occurs' in the glass of the bulb; especially pentodes of the series he;e described for the base between the fused-in leads. Conduction in glass, leads (filament, control" screen and suppressor like the electrolysis of liquids, is based upon the grids). movement of ions. In the èonduction in glass sub- The top leads (anode and a second connection of stances are also deposited at the poles. The presen~e the .supprcssor grid) have a considerably larger of these substances results in general in a lessened high-frequency current to withstand, since thc out- reliability' of the fused-in leads. The oxide layer .put circuit is connected to them. Of themselves usually present on the metal is discoloured, gas the bare molybdenum or tungsten leads, through bubbles are formed in the boundary layer and after glass would be quite satisfactory here; it is, however, some time the valve begins to leak. As "a result of difficult to fasten a suitable binding post arrange- di-electric losses'which always accompany the elec- ment to them. Direct connection is out of the trical conduction in the case of bulbs of transmitter question because of the brittleness of fused-in valves, this phenomenon becomes more disturbing molybdenum or tungsten. ' the shorter the wave length. A much stronger arrangement is obtained by the . Since all the ions which occur in glass are not use 'of heavy çopper pins which are fused to the equally mobile, by a suitable choice of composition bulb by means of a "fernico" ring soldered to them. of the glass a considerable improvement of the in- These leads can easily withstand 50 A, or more. sulating p0'Y'er can be attained. During recent In the trio des óf this' series copper with fernico is ,years types of glass have been successfully manufac- also used for the base connections (filament). , tured 'whose electrical conductivity is many th~usand times smaller than that of ordinary hard The zirconium getter glasses (fig. 5). Pure zirconium has a very strong tendency, de-

fO-5;11 pending on the temperature, to bind gases. When 6 / / used in powder form a large active surface is ob- 3 / / tained while gas is' then taken up even at low tem- . a peratures. In general it may be said that for bind- .... ~ fa 11 ing oxygen and nitrogen red heat is' necessary, 6 f'l' / / V while hydrogen is taken up only by the non-glowing 3 !J / / parts (see the article referred to in footnote 4). I~ In a transmitter valve all these, temperatures are la- 1/ / I 6 II / 1/ / '~a/ as a rule present at the same time. Thus in, the case . 17 , of the largest of the valves here described the 3 J 1/ / V/ maximum anode temperature is about 800 oe, the la,-I /I /r/ #VI lowest about 300 oe, while connection strips' and 100 200_ 400 C 500 J9ZJ8 the like may be still colde~. Since a good vacuum is Fig. 5. Electrolytic conductivity of glass 'as a function of the necessary here in the first place for maintaining temporature. In a piece of glass two wires of 1 mm diameter the cathode emission (tungsten-thorium cathodes are fused in over a length of 1 cm at a distanceapart of 1 cm. , Between the wires a D.C. voltage of 100 V is applied. The cur- are very sen,sitive to oxygen), it is obvious that the rent increases approximately according to a power of e with zirconium powder should be placed in the imme- the temperatüre. Four groups of glass can he distinguished: a) Borosilicate glass. Softening point 525-550 °C. Resistance diate vicinity of. the filament. ' , low. ( In' the case 'of the pentodes of this series the b) Soft glass (lead glass). Softening, point about 450°C. Resistance fairly high. ' screen grid as well as the anode is covered with c) Hard glass. Softening point 700-1000°C._Resistance high. zirconium. In the case of the triodes also it was d) "Electrolysis-free" glasses.' Softening point 550-600°C. Resistance very high. natura~ to make use ~f such a shielding cage which "

SEPTEMBER 1941' NEW .TRANSMITTER VALVES 257

, in this case could only be formed by the 'control siderably larger than the primary current, is reduced grid. While indeed the control grid of a to about the desired value. usually has a lower working temperature than the Covering the grid with zirconium would in such screen grid of a pentode, the 'getter action is in valves disturb the, equilibrium between the cur- this case increased by the fact that the zirconium rents, and thus 'have an unfavourable effect. In on the control grid has a low electric potential with order to give some idea of the change in grid cur- respect 'to the surroundings, so th~t any positive , rent caused hy covering it with zirconium, several ions present-are drawn to it. It was actually found characteristics are shown in jig. 6. In the case of possible. to maintain a. good vacuum in a triode the valve with zirconium-covered control grid, used' also by this means: in á high-frequenèy' connection, the grid current" was ~65 mA and the energy amplification 18 times, compared with 90 mA and 39 times in the case of . the ordinary valve. In order to prevent such an increase in the grid current, u;_ the new valves with .- zirconium getter only those parts. of the control grid were covered which ar.e not exposed to a direct bombardment by' primary electrons. The parts here involved are the grid rod and in some cases the half of.the wire 'surface which does not face the v fila~ent. In the case ofthe anode also covering (the inside) of the anode with zirconium aff~cts the charac- , . Fig. 6. Grid current characteristics of the valve TB 3/1 000, teristic. In jig. 7 the ia- Va characteristics of two a) without zirconium, b) with a very thin layer of zirconium powder on the grid. Because of the fact that the coefficient 1 000 W transmitter are shown. One has of secondary emissión of molybdenum is greater than unity' a: bare molybdenum anode; ~ the otller the' anode for electron velocities of several hundred volts, a negative grid current may occur in a). In practice, however, a high is covered internally with zirconium., In' the latt~r positive grid voltage is accompanied by a low anode voltage, so that with a well constructed valve the momentary value ,5;i! of the grid ,current .always remains positive, Ia !~2=500V Vg1=! Iiov .~~ t ; . The covering of electrodes with zirconium has A' I tOO still other resulta.. As was. described in the article referred to in footnote 3), zirconium decreases the, 3 / X .

of the surface upon which it is .N VL" 0 deposited. Since the total current to an- electrode 2 r r- is the sum of the primary eniission falling on the electrode and the secondary' emission counter to J;;:- ~ /r . it; influence can be exerted on the total current i by covering the electrode with ;more or, less. zir- K/ . I o I I-Va(V) , conium. 200 400 I 500 800 tOOO 150V In the case of the control grid the desire is as Ia ~ ; 4A L 1 a rule to make the total current as small as possible, I . Every increase in the grid current is accompanied Ij', I 100V I by an increase in the excitation power and thus 3 I 50V a decrease in the energy amplification. It is therefore ,r ---. __ best in this case when the secondary emission is- 2 r--, I as nearly as possible equal to the primary electron . I 0 current received. J.-- r . ! Now in the case of modern transmitter valves this I is usually approximately the case even without the o Ir- 1 -+- Va(Vj deposit of zirconium. They are constructed for low o 200 600 800 ' 1000 . 1200 voltage and great current density so th~t there is S94J4 so much in the valve that secondary Fig. 7. The ia- Va characteristics of two tetrodes .. The upper ,electrons pass through it with difficulty, and the characteristic refers to a with molybdenum anode; the lower characteristic is obtained when the anode is-covered secondary emission current, which of itself is' con- internally wit~ zirconium powder, 25R PHILIPS TECHr ICAL REVIEW Vol. 6, No. 9

cm -bO

- 55

-so

-40

-JO

-25

-20

riS

-10

Fig. 8. Two modern transmitter triodes (left) and two used formerly (right) of about the same output. The photograph illustrates the enormous decrease in external dimensions.

'ol case the familiar link which occurs as a result of I. the secondary emission for Vc,< V has disappeared 3000 g3 ~ almost entirely, and a characteristic is obtained ~ ~~-, ;:'000 , which may compare with that of an ideally con- ,, Tn V1 structed pentode 5). \ r---- \ Another favourable property of the powdered - t-, 1 tOOO zirconium is its great capacity for radiating heat, 800 -- f.Z6tn--.... which amounts to 80-90 per cent of that of a black wo --- ~ $,.,,_/8< ~o/~ ~. body 6). Because of this the specific dissipation of 500 "'I:r--,.~ the anode and grids could be further considerably .wo f'.,:::.._ increased. It is clear from fig. 8 how this permitted 300 -r..;Q:) a reduction in dimensions compared with the ~ ~ 1\,\I"'" -, 200 i--- earlier valves. r--.... \ ~ Thanks to the measures here described a grcat '\ improvement could be obtained in the short wave 100 1'\\ 1\ properties. Most of these transmitting valves can be 8(} -\ used on wave lengths down to 3 m without the ef- \ -~ 60 \ ficiency, which amounts to 70 to 75 per cent for 50 1 long waves, being too much diminished (seefig. 9). 40

30

20

5) In judging this characteristic it must be kept in mind that ~ ISm different requirements are made of a transmitter valve ID 107 2 4 6 8' f08 2,f(Jfper/ sec than of a low-frequency output amplifier valve. Thus the J924/ "zirconium tetrode" when used as low-frequency amplifier will still give too much distortion, while in excitation or Fig. 9. Survey of the maximum obtainable outputs of the new amplification of high-frequency energy this is of no im- series of transmitter valves on short waves. In different cases portance. the triodes give a higher output than the corresponding pen- 6) Other finely divided substauces such as tungsten also pos- todes; this takes place, however, at the expense of a much sess this property (see the article referred to in footnote 3). higher excitation power.