these m agnetrons are essentially identical. Geometrical and performance d a ta are given in Table I I I . As m ay be seen, these m agnetrons represent an increase in power capabilities by a sizable factor over the 725A. The 4J52 m agnetron has been used extensively under the severe conditions of 5 ¿us pulse duration, a repetition rate of 200 pulses per second, 15 kv. and 15 amps, input. U nder these conditions it has n o t been possible to elimi
nate entirely a tendency to arc, although good perform ance is obtained.
Fig. 74— T he 4JS2 “ packaged” m agnetron (100 kw ., 9375 m c/s).
In m ost respects these m agnetrons represent the optim um achieved in m agnetron design for wavelengths near 3 cm. during the war. Largely because of the lack of ham pering electrical and mechanical restrictions, it was possible to make use of all the latest and best techniques in the design of each of its parts. T hus each possesses the desirable features of “ p a c k aging,” axial cathode m ount, wave guide o u tp u t, and high efficiency a t low pulling figure.
2 0 . Ma g n e t r o n s f o r Wa v e l e n g t h s n e a r 1 Ce n t i m e t e r
20.1 Preliminary Work: Intensive work on m agnetrons for wavelengths near 1 cm. began when a joint program w ith the R adiation L aboratory a t
M A G N E T R O N /1.S G E N E R A T O R OF C E N T I M E T E R W A V E S 333
Columbia U niversity was undertaken to prepare for m anufacture a variation of a m agnetron developed there. Prior to this, some work prim arily of an exploratory nature had been done in our Laboratories. W hen it h ad been dem onstrated th a t scaling from the 10 cm. range yielded an efficient 3 cm.
magnetron, similar a ttem p ts were made in scaling to wavelengths of 1 to 2 cm. M agnetrons of such wavelengths, obtained by scaling from eight
Fig. 75—A view of a cut-aw ay 4J52 “ p ackaged” m agnetron (100 kw ., 9375 m c/s) which in in tern al d etails is essentially like the 4J50 m agnetron (280 kw., 9375 m c/s).
Of special in terest arc: the axial cath o d e construction including perm cndur end pieces and radiative extension; th e q u a rte r w avelength transform er of Ii-sh ap e cross section (between the o u tp u t resonator and the line w ave guide); an d the w ave guide w indow con
struction w ith-associated chokes.
resonator, 10 cm. models like the 706A-C, were unstrapped while others were strapped with the early British type of strapping and w ith double ring type strapping. > O thers were made by scaling from twelve resonator, 3 cm.
models and were strapped w ith both double and single ring straps. T he output circuits used were of the loop and coaxial type.
Although some of these m agnetrons were faulty, prim arily in their o u tp u t circuits, the m ajority oscillated. One m agnetron of the unstrapped eight resonator variety, the most successful type, was operated over a consider
able period in our Laboratories. Although no m agnetron was developed concentrated on m agnetrons of very short wavelength. The first Columbia m agnetron th a t was reasonably successful was unstrapped, having a vane
20.2. The 3J21 Magnetron: The Bell Laboratories undertook, in collabo
ration w ith the Columbia Laboratory, to design a “ packaged” version of the 3J31 m agnetron for m anufacture by the W estern Electric Co. A t the time, it had not been decided w hether the new m agnetron should have a strapped or a “ rising sun” resonator system. Considerations having prim arily to do w ith m anufacturing techniques indicated the latter to be preferable although the form er would have been possible. Accordingly, work was started on a “ packaged” , “ rising sun” m agnetron to oscillate at
M A G N E T R O N A S G E N E R A T O R OR C E N T I M E T E R W AVES 335
In the development of the 3J21, the “ rising sun” resonator system used by the Columbia L aboratory was adopted practically w ithout change. The ratio of the n atural frequencies of the large and small resonators is approxi
m ately 1.8. T he anode diam eter is 0.160 in. and the anode length 0.190 in.
The structure is fabricated by the so-called “ hubbing” technique,20 which had been brought to a high sta te of developm ent a t the Columbia Laboratory for the purpose. In this technique, a hardened steel die or hub, machined
The problem of the dissipation of the considerable heat of back bom bard
ment was complicated by the necessity of activating the cathode a t a temperature of 1050°C. The requirem ents of heat dissipation and activation oppose one another, one calling for low therm al impedance of the cathode
*8The technique is here called “ hubbing” rath er th an hobbing for two reasons: T he term hobbing has a m eaning in m achine practice quite a p a rt from the technique in ques
tion. T he term hubbing is used for processes in which the interior of a piece is removed by forcing a m ember into it. I t presum ably arose from the early practice of m aking axle holes by forcing a hardened steel m em ber through the w rought iron wheel hub. F u rth e r
more, it has an analogous usage in coining.
support, the other, high. T o meet each requirem ent satisfactorily with considerably larger and more rugged than if placed inside the cathode, whose diam eter is 0.096 in. A view of the 3J21 m agnetron cathode structure is rod considerably beyond the active surface, as seen in Fig. 78. T he cathode should also have good therm al conducting properties in this extension, conductivity and stru ctu ral strength. Copper, silver, and nickel do n o t meet all of these requirem ents. T he first cathodes of promise were turned from
M A G N E T R O N A S G E N E R A T O R OF C E N T I M E T E R W A V E S 337
a heavy central support lead inside the K ovar cone through a hole pro vided in the cone. The input end of the heater cham ber is flared to ac
commodate the K ovar support cone to which it is brazed. All of the m a
terials are thus of good stru ctu ral strength. The tungsten and m olybdenum are good therm al conductors, the K ovar not. Some p a rts of the cathode were grit blasted to increase their radiative emissivity.
The to tal axial motion of the cathode by therm al expansion from cold carried the choke joints and wave guide window seal. T his la tte r piece was fabricated by hubbing a rectangular w ave guide in a copper cylinder into easily be controlled. A rigid inspection of both electrical and mechanical
properties of each m agnetron before sealing and pum ping was an absolute necessity to insure reproducibility. T he use of an iris coupling between the resonator and the wave guide presented a possible solution. However, the iris required would be too large to be placed directly a t the back of the resonator. It was proposed to p u t a hubbed rectangular resonant iris a t the back of the resonator and a decoupling iris a half wave length distant in the wave guide. This construction provided a resonant cavity between the two irises in which the storage of energy would provide some degree of frequency stabilization. T he structure of this o u tp u t circuit m ay be seen in the photograph of a cutaw ay model in Fig. 77. B oth higher efficiency of
Fig. 76—An external view of the 3J21 “ packaged” m agnetron (60 kw., 24,000 m c/s).
operation and lower pulling figure th an obtained w ith the earlier circuits resulted. As average values, an increase in over-all efficiency from 24 to 28 per cent and a drop in pulling figure from 25 to 18 m c/s m ay be cited.
T he R F voltage breakdown strength of the wave guide window was marginal. A new design developed a t the Columbia and M. I. T. L abora
tories, incorporating a larger window and “ stream lined” wave guide con
tours adjacent to it, was modified slightly and used in the 3J21.
The magnetic pole gap was made 0.290 in. T his was as small as was felt practical. T his and other steps were taken in an effort to save as much as possible on m agnet weight.
The mechanical problem s connected w ith the fabrication of the 3J21
M A G N E T R O N 4 5 G E N E R A T O R OF C E N T I M E T E R W A V E S 339
m agnetron resulted largely from the small tolerances it was found necessary to impose. T he tolerance allowed on radial location of the cathode was
± 0.002 in .; on transform er slot w idth (when this o u tp u t was used), ±0.0005 in.; and on dimensions of the anode structure, ± 0.001 in. Deform ations, such as those produced by placing a m agnetron with slightly misaligned pole pieces on a strong m agnet, had to be carefully avoided. T he im portance
Fig. 77—An in tern al view of the 3J21 “packaged” m agnetron (£0 kw., 24,000 m c/s).
Note the “ rising su n ” resonator system , the stabilizing cavity in the wave guide o u tp u t between a rectangular resonant iris a t the “ back” of the o u tp u t resonator and a circular decoupling iris a half w avelength d istan t, and the beveling of the w ave guide edges ad jacen t to the wave guide window.
of cathode alignm ent, for example, m ay be seen from the fact th a t across the anode-cathode space, which in the 3J21 is only 0.032 in., there is an operat
ing voltage gradient of 165 k v ./in . T his gradient is doubled in the region where the cathode support passes through the hole in the pole piece, at which point there is a nominal clearance of 0.015 in. T hus the elimination of small burrs and small radii was essential.
An external view of the 3J21 m agnetron is shown in Fig. 76, an internal view in Fig. 77. O perational and other d ata are to be found in Ta b l e IV.
T h e p r in c ip a l o p e r a tio n a l p ro b le m s e n c o u n te re d in th e 3 J2 1 m a g n e tro n the cathode. Cleaning and plating solutions release atom ic hydrogen which readily perm eates the iron pole pieces and other parts. Various
M A G N E T R O N A S G E N E R A T O R OF C E N T I M E T E R W A V E S 341 showed th a t the only observable causes were resonator distortion, the presence of brazing flux or other foreign m aterial, and off-center cathode location.
An interesting phenomenon, discovered during the developm ent of the 3J21 m agnetron, is th a t known as the “cold s ta rt.” W hereas m ost magne
presents a lower limit on pulse duration which m ay be employed and further
servatively, the 3J21 cathode under extrem ely severe conditions.
L ittle difficulty with cathodes has been experienced in pulsed magne the high emission density required (approxim ately 30 am p s./cm .2) the active coating was rapidly lost as a result of arcs, and frequently the nickel support itself was fused and vaporized. is subjected. Increase in power input, either as increased voltage or current, or both, rapidly increases the frequency of arcs. Similarly, increased pulse
Fig. 78—A group of cathode stru ctu res for centim eter w ave m agnetrons. From left to right th ey are: (1) the SJ26 cath o d e—oxide c o at
ing on roughened nickel base; (2) the 4J21-30 cathode—oxide coating on roughened nickel base; (3) th e 720A-E cathode— m etallized, oxide coating on nickel w ire mesh base; (4) the 706AY-GY cath o d e—oxide coating on sm ooth nickel base; (5) the 4J50 cathode an d su p p o rt stru c ture—oxide coating on sintered, nickel powder, m atrix base; (6) the 725A cathode and su p p o rt leads—oxide coating on sintered, nickel powder, m atrix base; an d (7) the 3J21 cathode and su p p o rt stru c tu re —oxide coating on sintered, nickel pow der, m atrix base. T h e active cathode portion in each of the stru ctu re s (5), (6), an d (7) is indicated by a bracket.
MAGNETRONAS GENERATOR Oh' CENTIMETERWAVES
length m akes the arcing much worse; in the 725A for example, changing prim aiy and secondary electrons. All-metal, secondary em itting cathodes with prim ary em itting, sta rte r cathodes have been tried in some labora
tories with some success b u t have not y et come into general use. The efforts a t the Bell Laboratories have been directed tow ard developing an d im
proving the oxide coated cathode along lines m aking it more nearly possible to satisfy requirem ents (3), (4), (5), and (6) listed in Section 10.7 Magne
tron Cathodes of P A R T I w ithout im pairing the ability to m eet require
m ents (1) and (2).
M ost of the cathode developm ents were made in connection w ith the 725A.
T his m agnetron served as a convenient “ laboratory” or “proving ground” activation, by controlling the initial break-in of the operating m agnetron, and by devising m eans of q u antitatively determ ining some measure of the adequacy of any given cathode. The last of these item s involved the de
velopm ent of an autom atic counter which registers the num ber of arcs or bursts of arcs which cause the current to exceed a predeterm ined value. By recording the accum ulated num ber of arcs a t ihtervals throughout the life of the m agnetron, it is possible to get a picture of the arcing p a tte rn with life. Such counters have done much to p u t the life testing and initial break- in of m agnetrons on a sem i-quantitative basis. The smoothed curves shown in Fig. 80 were obtained through the use of these counters.
E arly atte m p ts a t building a cathode upon which sufficient active m aterial m ay be held, made a t both the M. I. T. R adiation L aboratory and the Bell Laboratories, involved winding coated cathodes w ith nickel wire to provide a reservoir of active m aterial under the wires. In this m anner, the m aterial was protected from direct arcing effects, allowing the active m aterial to m igrate slowly to the outside surface of the wires. T his general idea was greatly improved upon when the wire winding was replaced by a woven nickel mesh into the interstices of which the active m aterial was packed. The
M A G N E T R O N /I S G E N E R A T O R OF C E N T I M E T E R W A V E S 345
nickel mesh, being spot welded or sintered directly to the nickel base, formed a p a rt of the cathode foundation, helping to reduce the coating resistance.
I t was found th a t arcs, rath er th an striking directly to the active m aterial, tended to strike to the nickel wires. In spite of this fact, sudden bursts of current in arcing would eru p t some of the active m aterial from the cathode.
The mesh used in the 725A cathode is made of 6 mil nickel wire woven with 75 wires to the inch. T he radial thickness of the mesh on the cathode is 10 mils. The average am ount of active double carbonate m ixture which the cathode holds is 23 mg. per sq. cm. The mesh cathode is shown in Fig.
79 in three stages of its fabrication. W ith this cathode, it was possible to
Fig. 79—725A cathode construction. Shown are: (1) the heater elem ent; (2) the machined nickel cathode blank; (3) the cathode blank w ith the nickel wire mesh, welded or sintered in place; (4) the mesh type cathode with oxide coating applied, c u t aw ay to show heater elem ent in place; (5) the more recent cathode w ith sintered, nickel powder, matrix base (uncoated).
produce 725A m agnetrons having a guaranteed life a t rated operating conditions of a t least 500 hours, the average value being considerably higher.
Fhe arcing characteristic of the mesh cathode, excluding the initial break-in Iienod, is shown in Fig. 80. The abrupt failure is typical of magnetron cathodes. A t shorter pulse lengths failure occurs considerably later b u t in the same ab ru p t fashion. T h e life curve of the early type of cathode con
sisting of an oxide coated nickel cylinder, if plotted with the d a ta on other cathodes in Fig. 80, would be crowded very close to the axis of zero hours.
Even w ith cathodes of mesh construction, operation of the m agnetron was generally inadequate a t 5 ¡is pulse duration. Furtherm ore, consider
able break-in tim e was necessary before stable operation a t 1 fis pulse d u ra
tion was achieved. Some of the first mesh cathodes used in the 725A,
for example, took as long as 45 m inutes of intense arcing during gradual increase of input voltage and current before satisfactory operation was attained. I t seemed unlikely th a t the arcing during the break-in period could be completely elim inated, b u t it was found th a t im provem ents in cathode construction which made steady operation a t higher current possible also reduced the tim e of initial break-in to atta in these conditions.
I t appeared th a t the resistance of the cathode coating m ight be lowered by distributing nickel more uniformly throughout the coating. Accord
ing. 80—C urves of arcing during life for 725A m agnetrons, h aving various types o cathodes, o p e ra tin g a t 5.7jrspulse d u ration, 165 pulses per second, and 210 kw. p e a k p o w e input. I t is to be em phasized th a t these conditions arc considerably more strin g en t than the norm al operating conditions. C urve (a) is for a cathode having oxide coating on a nickel wire mesh base; curve (b), m etallized coating on a nickel wire m esh base; curve (c), eith er plain oxide or m etallized coating on a sintered, nickel powder, m atrix base.
ingly, mesh cathodes were coated with a m ixture of double carbonates in which was distributed a fine nickel powder of high p u rity having an average particle size of 2 microns. T he carbonate particle size is between 1 and 2 microns. The am ount, found not to be critical, was 55 per cent by weight of the combined dry ingredients of the coating mixture. Although this reduced the am ount of active m aterial present on the cathode surface, as seen in Fig. 80 it resulted in a cathode having considerably improved arcing characteristics w ith considerably longer useful life. T his cathode con
struction was adopted finally for the 725A production and was also utilized in the 10 cm. high power m agnetrons, the 4J45-47, for operation a t 5 ¿is pulse duration.
M A G N E T R O N A S G E N E R A T O R OF C E N T I M E T E R W A V E S 347
viding considerable radiating area imm ediately adjacent to the cathode surface by extending the cathode, as shown in Figs. 75 and 77, into the pole I echnology and a t Columbia U niversity generously supplied personnel for several cooperative projects and information on new results. In the 725A our Laboratories and consulted generally on m agnetron problems.
Mechanical design, as well as constructional work and preparation of the detailed specifications for m anufacture, was carried out by V. L. Ronci, D. P. B arry, F. FI. Best, J. E. Clark, D. A. S. Hale, J. P. Laico and their associates, including T . Aam odt, C. J. Altio, D . I. Baker, C. Blazier, R. H.
Griest, F. B. Henderson, W. Knoop, W . J. Leveridge, J. B. L ittle, C. Maggs, j . A. Miller, H . W. Söderström , and F. W. Stubner.
Perm anent magnets were designed by P. P. Ciofli of the m agnetics group in the Physical Research D epartm ent. N um erous physico-chemical prob
lems were solved by L. A. W ooten and his associates of the Chemical De
partm ent. The work of J. B. Johnson and J. R. Pierce in allied fields contributed to th a t described in this paper. In addition, m any users of m agnetrons for rad ar purposes throughout the Laboratories collaborated by conducting tests of m agnetrons used in rad a r equipm ents. We wish to acknowledge the support and advice of M . J. Kelly and J. R. Wilson under whose general direction the work proceeded.
Finally, the authors wish to thank L. M . Field, W. B. H ebenstreit, G. E.
Moore, A. T. Nordsieck, and N . W ax, who have read p a rts of the paper critically.
C on trib u tors to th is I s s u e
J. B. Fi s k, S.B. in A eronautical Engineering, M assachusetts In stitu te of Technology, 1931; Redfiekl P roctor Traveling Fellow, T rin ity College, Cambridge, 1932-34; P h.D . in Physics, M .I.T ., 1935; Ju n io r Fellow, H a r
vard U niversity, 1936-38; Associate Professor of Physics, U niversity of N orth Carolina, 1938-39; Bell Telephone Laboratories 1939-. Engaged in magnetron work in the Electronics Research D epartm ent during the war;
now in the Physical Research D epartm ent.
Ho m e r D . Ha g s t r u m, B .E .E ., 1935; B.A., 1936; M .S., 1939; P h.D . in Physics, 1940, U niversity of M innesota; Research and Teaching A ssistant- ship, University of M innesota, 1936-40; Bell Telephone Laboratories, 1940— ; engaged in the developm ent of m agnetrons during the w ar in the Electronics Research D epartm ent; now in the Physical Research D epartm ent.
Pa u l L. Ha r t m a n, B.S. in E .E ., U niversity of N evada, 1934; P h.D . in Physics, Cornell University, 1938; In stru cto r in Physics, Cornell U niversity, 1938-9; Bell Telephone Laboratories 1939-; has been a m em ber of the Electronics Research D epartm ent, being engaged during the w ar in the development of m agnetrons; now in the Physical Research D epartm ent.
349