1955

Warnecke, Guenard, Doehler, and Epsztein: The " M"-Type Carcinotron Tube

413

practical considerations, such as available production facilities and engineering costs, also frequently exclude their use. These factors, however, will become less important as more engineering know-how and production experience are gained. The relative simplicity of the magnetic amplifier makes for easier, cheaper and faster assembly once a system has been properly designed and

engineered. The magnetic amplifier is no longer a laboratory device but has been tested and proven in actual military and industrial service until today it is a major factor in the design and development of an ever-increasing number of new and improved systems. Reliability is by no means the only virtue of the magnetic amplifier. In many applications it has proven to be superior in performance, size and cost to vacuum tube

amplifiers, rotary amplifiers and even mechanical devices. No other amplifier offers such a combination of long life; low maintenance; ruggedness; resistance to extreme conditions of vibration, shock and temperature; no warm-up time; and high efficiency, in a completely static device which may be permanently sealed or potted. The day of the magnetic amplifier has arrived.
ACKNOWLEDGMENT The author gratefully acknowledges the assistance of the many people in the magnetic amplifier field who so freely gave their time and advice in the initial preparation of this paper. Special thanks are given to E. T. Hooper and W. A. Geyger of the Naval Ordnance Laboratory for assistance in the final preparation of the manuscript.

The

"M"-Type
O. DOEHLERt,

Carcinotron Tube*
AND

R. R. WARNECKEt, FELLOW, IRE, P.

Summary-This paper presents theoretical and experimental results concerning the "M Carcinotron." In particular, the influence of space charge has been considered, thereby permitting an explanation for the measured values of starting current, the influence of the coupling impedance on efficiency and the existence of parasitic oscillations. The "rising sun effect" which should be present in these tubes, as it is in the magnetron, has been investigated theoretically and experimentally. The experimental results exhibit a decrease of efficiency in the predicted range of operation.

GUtNARDt,

B. EPSZTEINt

SENIOR MEMBER, IRE,

INTRODUCTION T HE Carcinotron' tubes are backward-wave oscillators. Their structure is characterized by the following features: 1. An electron beam is in interaction with a backward space harmonic of a delay line. 2. The power output is located at the gun end of the interaction space. 2 P. Gu6nard, 0. Doehler, B. Epsztein, and R. Warnecke, 3. Means for absorbing rf energy reflected by possible "Nouveaux tubes oscillateurs A large bande d'accord 6lectronique output mismatch are introduced in the rf field of the pour hyperfr6quences," C. R. Acad. Sci. (Paris), vol. 235, pp. 235delay line near the collector end generally inside the 236; July, 1952. They were given previously, together with experimental results, by Epsztein at the 10th Conference on Electron Tube tube. Pesearch, Ottawa, Can.,june 1952, in a discussion on R. Kompfner's This structure gives a very wide electronic tuning paper "Packward waves-," presented at this conference. R R. Warnecke and Guenard, "Some recent work in on range, and frequency insensitivity to load impedance. 'new types of valves forP.highest radio-frequencies," Proc. Francevol. IEE, Two types of Carcinotron tubes have been investi- 100, part III, p. 351; November, 1953. 4 R. Warnecke, P. Gu6nard, gated: the "O" type, in which the beam travels in an mentaux dans- les tuibes a ondeand 0. Doehler, "Ph6nombnes fondaprogressive," L'Onde Electrique, no. interaction space at constant dc potential, as in the 325;5 April, 1954. R. Warnecke, 0. Doehler, and B. classical traveling-wave tube, and the "M" type, where wide P. Gu6nard,tuning high efficiency microwave Epsztein, "A new oscillator, the M electronic the beam travels perpendicularly to crossed electric and Carcinotron,"' paper presented at the 11th Conference on Electron Tube Research, Stanford University, Stanford, Calif., June 1953. magnetic fields, as in the magnetron amplifier. 6 P. Gu6nard, "On some results obtained with 0 and M type Carcinotron," paper presented at the 12th Conference on Electron * Original manuscript received by the IRE, November 9, 1954; Tube Research, University of Maine, Orono, Me., June, 1954. 7 0. Doehler, "Space charge effects in traveling wave tubes using revised manuscript received December 16, 1954. f Electronics Dept., Center of Tech. Res., Compagnie G6n6rale de crossed E-H fields," paper presented at the Symposium on Modern Advances in Microwave Techniques, New York, N. Y., October, T6l6graphie sans fil, Paris, France. 1 Registered trade-mark of the Compagnie G6n6rale de T.S.F. 1954.

This paper deals with the "M" Carcinotron. The structure of this tube and the results of a simplified small signal theory were given in a short note published in 1952.2 The methods used to obtain these results were more fully described later.3 4 The aim of this paper is to describe with more detail the properties of the "M" type Carcinotron, gathering and completing the information given previously.5-7 A small signal theory taking into account spacecharge effects is established, allowing expressions for the starting current and frequency, build-up time and frequency pulling to be derived. The results of this theory are checked against experimental data. Some effects typical of the "M" Carcinotron (rising sun effect, parasitic oscillations) are explained. The practical interest of the "M" Carcinotron is best shown by the per-

414

PROCEEDINGS OF THE IRE

A pril

formances obtained on this type of tube, some of which are given at the end of the paper. These performances, together with those previously published,3'4'8 suggest that the "M" Carcinotron should be best suitable when high power high efficiency operation is required.
DESCRIPTION OF THE "M" CARCINOTRON Fig. 1 shows a linear version of the "M" Carcinotron. A delay line L is perfectly matched at the end M near the collector K by means of an attenuating material to avoid reflections. N is the output. An electron beam F, produced by the gun, travels parallel to the delay line L, and the sole S in the x-direction under the influence of a constant and uniform magnetic field B in the z-direction and an electric field Eo in the y-direction due to the voltage VO applied between L and S.

longitudinal electric field. The electrons which transfer energy to the rf field, approach the anode and are brought to a higher dc potential, their velocity remaining approximately constant and equal to Eo/B. As compared to the mechanism of operation of the "O" Carcinotron, it can be said that in the travelingwave tubes with crossed electric and magnetic fields the potential energy of the electrons is transformed into rf energy; while in the "O" type traveling-wave tubes the kinetic energy of the electrons is transformed into rf energy.
SMALL SIGNAL THEORY Small Signal Theory Neglecting Space Charge In this theory, a rectilinear beam is considered. In

Vp
Fig. 1-Schematic structure of a linear "M" Carcinotron.

Interaction between the beam and the backward space harmonic will occur, if the phase velocity of the space harmonic is equal or nearly equal to the mean electron velocity Ve = Eo/B, i.e.: 11 -' V's = Eo/B. (1) The positive feedback introduced by the beam leads to an oscillation if the current is high enough. This oscillation occurs with perfect matching at both ends of the line, the necessary feedback being furnished by the beam only. The oscillation frequency will be that for which (1) is fulfilled. The beam focusing and the transfer of energy from the beam to the rf field occur in the same manner as in the magnetron amplifier:9 the transverse rf field produces a beam bunching in the favorable phase of the longitudinal rf electric field. The energy transfer from the beam to the electromagnetic field of the line is achieved through the
R. Warnecke, "Sur quelques resultats r6cemment obtenus dans le domaine des tubes pour hyperfr6quences," Ann. Radioelect., vol. ix, pp. 107-1l35; April, 1954. 9 R. Warnecke, W. Kleen, A. Lerbs, 0. Doehler, and H. Huber, "The magnetron type traveling-wave amplifier tube," PROC. I.R.E., vol. 38, pp. 486-495; May, 1950. See also J. R. Pierce, "Travelingwave Tubes," D. Van Nostrand Co., Inc., New York, N. Y., chap.
8

Line

the equations of the static trajectories X = VeT + a cos (&)rT + k) y = yo + a sin (w,r + p), where r is the transit time and co,= eBIm, the cyclotron angular frequency, the amplitude a of the rolling circle is assumed zero, a condition which can be obtained with a proper set of initial conditions. Assuming that all rf quantities vary as Ei(wt-rx), there exist, in the absence of coupling between the beam and the line, six waves Ax and by arbitrary r, = 25 = Fe = Ax = 0, by = 0 Beam co + co, by= + jax r3,4 = aX = T jwrSx, jQ,6y y + Ve E= jKEx I r2 = rO
=

and K = coth ryo. The coupling between the beam and the line can modify significantly only those waves for which the propagation factors are near one another, i.e. the two beam waves Fi, r5 and the line wave F2. This assumes that one space harmonic only is considered. There is a possibility of simultaneous coupling of beam waves with different space harmonics, a question which will be discussed later (rising sun effect). The theory (Appendix A) shows that one of the beam waves, say r5, is not coupled to the line. For this wave: x = jK8y Ax = = 0 r5 = r, Ex = E, = 0. The modified values of r1 and r2 are solutions of the

r6 = - rO Ey = -jKEZ, where 8z, by, Ax, by are the rf components of electron motion, E_ and EI, the field components in the beam,

equation:

(r - r)(r

r)

r02rRIoK =YM2 Eo

(2)

XV; 1950.

where R, is the coupling impedance, Io and Eo the dc beam current and electric field. The two solutions can

1955 be written:

Warnecke, Guenard, Doehler, and Epsztein: The "M"-Type Carcinotron Tube

415

Vo = V' (5) Small Signal Theory Taking Into A ccount Space Charge 7r Ml-= +2rn(n=0,1,2...)(6) It is no more possible in this case to consider an in2 finitely thin beam. As shown by Brillouin,'0 linear traEq. (5) says that for every possible oscillation (n = 0, jectories are possible, assuming an equipotential if the plasma frequency Q0 equals the cyclotron 1, etc.) the phase velocity of the line at the oscillating cathode, frequency is equal to the electron velocity. Eq. (6) is the frequency W.: condition for the starting current.2 e po== eB QO /{/ Eo =O m = 2n) (7) 02 e2R tangh ryo, the charge density in the electron beam. The po being I being the length of the line. The influence of an attenelectron velocity varies inside the beam as wry. If a nonequipotential cathode is used, it is possible uated line and of reflections already have been discussed.4 to avoid this condition; with a constant charge density If space charge effects are taken into account, (3) must be used, and the starting conditions are: in the beam, the electron velocity varies as yQo2/co. It has been supposed, with no other basis than the a rd = agreement between theory and experimental data, that (8) 2 it is possible to apply the results thus obtained to a beam u R. Warnecke, 0. Doehler and 0. Bobot, 'Les effets de la charge 10 L. Brillouin, "Trajectories in a single anode magnetron," Elec. d'espace dans les tubes & propagation d'onde a champ magn6tique," Ann. vol. 1946.

with complicated trajectories, if the charge density in the linear beam is taken equal to the average charge density rl,2= -m vFrd + yM2, r in the actual beam. The general space-charge theory leads to a tranwhere scendental equation which is difficult to solve. Therero+ r. rm = fore, the influence of space charge has been introduced 2 as a perturbation in the theory without space charge re - ro in the following manner:" The beam travels in an rf Fd = 2 field which is the sum of the field 4L guided by the line and the field J?, created by the space charge. The For these waves: trajectories of the electrons are determined by 4L+cIC, K E, Ax while the transfer of energy from the beam to the line A ~~- b5y=j -Y ~=I-(F'5x = Fe)Ve3X, 6QVi is determined by 1L only. The approximation consists K K r-r,e Eo in calculating Jb. from trajectories determined without The three other waves are only slightly modified by the space charge. This leads (see Appendix A) to the followcoupling. ing values of the propagation constants for the two This theory can be extended to a beam of finite thick- principal waves: ness if the trajectories are linear, which means that a nonequipotential cathode is used. It is then found that Ti,2=rm rd + M 2 (3) rd + \rd2+7M ) the rf charge density inside the beam is zero (see Appendix A). In addition to assumptions already mentioned, it has where been supposed that the periodicity of the line structure 002ra az = has no influence on the static trajectories, i.e. that the WVe distance between the line and the sole is large as compared to the pitch of the line. A two-dimensional prob- 2A is the width of the beam. lem has been treated, which supposes the structure inStarting Conditions finite in the z direction. The boundary conditions give the starting conditions These six waves make it possible to satisfy the boundary conditions, i.e. the values of 8x, by, Ax, by at for oscillations4 x = 0 (gun end of the line) and the existence of reflection (2-r e)(aoaieirFu- -ir2i) factors ao and a, at the ends of the line. If it is supposed = (r -rF)(aoatelTol -IT1,P). (4) that Ax = Sy = A = by = 0 for x = 0, the amplitudes of the waves r3, r4, rF are found equal to zero and there re- If the line is matched at least at one end (aoaz = 0) and if main only the two principal waves rF, r2 and the re- it has no attenuation, the starting conditions are: flected wave r6.

/0

Commun.,

p.

460;

Radioelect.,

V, p. 279; October, 1950.

416
d

PROCEEDINGS OF THE IRE

22 2

A pril

I
1It

\
Fig. 3-Starting current as function of voltage. (1) Without space charge-linear trajectories. (2) Without space charge-cycloidal trajectories. (3) With space charge-linear trajectories. (4) With space charge-cycloidal trajectories. (5) Experimental curve.

3. The increase of electromagnetic energy dW stored in the section dx during the time dt. dW = j(C, coo)Wdxdt
-

.6

Fig. 2-Build-up time as a function of beam current.

W=stored energy per unit length, =o=angular frequency of the free wave.

'YM

I+

3a2

YM2=

-+2

7r

If space charge is neglected, the equation which determines the propagation constant then has the form:
(9)

7rn.

r=

CO - COO

M2

(10)

1r-c Eq. (8) shows that, in the presence of space charge, ve the phase velocity of the line at the oscillating frequency is not the same as the electron velocity and asv,' is the group velocity. sumes different values for the various values of n. The boundary conditions of rf current for x = 0 and of Eq. (9) shows that the starting current is decreased electric field for x =I permit determination of co. by the effect of space charge. If there is a perfect match at least at one end of the tube and if the line has no attenuation, and r are Build- Up Time given by: If the beam current is higher than the starting cur=CO (11) rent, the tube oscillates and the rf amplitude starts in2 creasing with t, the time dependent factor being:
w

cos

[2YM'I/1

(-)2]

e jxt

aj(Ot+tIT
=

The time r characterizes the build-up time for small values of t and permits the determination of the approximate order of magnitude of the build-up-time. The calculation of r is analogous to the calculation of the starting current. But the balance of power must be modified to take into account the increase of the amplitude. The power transferred by the beam to the rf field in a section dx of the line, and during the time dt, is the sum of: 1. The energy absorbed by the line in the length dx and during the time dt. 2. The increase of power, propagated along the line in the section dx and during the time dt.

[1

2_

(12)

with:
@
=

r-_F7
|V

|Ve

(13)
0,

Eq. (12) is a transcendental equation for the current, is given. In Fig. 2,

1

if

yM,

i.e.

IVI

has been plotted

as a

function of I/II (I.T= starting cur-

1955

Warnecke, Guenard, Doehler, and Epsztein: The "M"-Type Carcinotron Tube

I
mA
- I

417
--

~~~~~~~~~~~~~~-

B= 1060
V5.- 525
800

G
V

I__//

Fig. 4-Frequency of parasitic oscillations as a function of voltage (solid curve-normal oscillation).

rent). If I/I, is large, r is given by the approximate

pression:
= r

ex-

X.\/I/I

1 _*I

i vg

1U

1
Yolts Fig. 5-Theoretical starting current for normal oscillation and first parasitic oscillation.

In practice, T is of the order of 10-9 seconds in the "S" band tubes. The total build-up time is then of the order of 10-7 to 10-8 seconds. In the presence of space charge, r decreases, so that (12) and (13) give the upper limit for r.

In Fig. 4 the main frequency and the measured sidebands as a function of voltages are shown. The difference of frequency between successive bands is approximately 60 mc. The amplitude is smaller for sidebands of EXPERIMENTAL RESULTS higher order. Starting Current These oscillations can be attributed to the excitation The theory for the starting current has been checked of the higher orders as given by (9). In Fig. 5 the on different types of experimental tubes with a linear theoretical starting current of the zero order (n = 0) and structure. Fig. 3 shows the results obtained with a of the first order (n =1) have been plotted. It follows linear tube. The starting current is a function of the that the starting current of the first order for low shape of the trajectories, and it has been calculated voltages is only three times higher than that of the zero for two different electron guns. For the "ideal" gun tra- order. The frequency difference between these two osjectories are straight lines; for the "magnetron gun tra- cillations is almost independent of voltage and, for the jectories are cycloids as in the plane magnetron without case of Fig. 4, has from (8) a theoretical value of 45 to space charge. In practice, the trajectories are between 70 mc for different trajectories of the electrons. these two limits. Curves 1 and 2 have been calculated The high frequency sidebands can be explained by infrom (6) neglecting space charge. Curves 3 and 4 have termodulation between these two oscillations, and an been calculated from (9). Curve 5 has been measured. asymmetry must appear, as shown in Fig. 4. The theory neglecting space charge gives starting currents much too high, especially at low voltages; the Pushing Figure space charge has an important influence on the starting If small signal theory with space charge were used conditions and the theory gives the correct order of to calculate the frequency as a function of current, large magnitude. variations (a few per cent) of frequency should occur. But measurements show that the pushing figure is relaParasitic Oscillations tively low. Measurements on a tube in the "S" band For high currents and low voltages, parasitic oscilla- have shown that variation of frequency is 2 to 3 mc tions can be observed. Sidebands with an amplitude for high voltages (5,000 v) and 5 mc for low voltages from 1/100 to i that of the principal frequency occur. (2,500 v) when the power output varies from one to ten.

418
p

PROCEEDINGS OF THE IRE

A prdl

1000

I)

w )

800

600

400

200

Fig. 8-Oscillogram of power output against control electrode voltage.

200

400

60O

lo (mA)

800

Fig. 6-Power output as function of beam current for various frequencies.

'1

Fig. 9-Power output and beam current as a function of plate voltage.

For some frequencies, especially at low voltages, P =f(I) has a more complicated form. Up to now, this has not been explained. Fig. 7 shows rf power and beam current as a function of control grid voltage for two frequencies. Fig. 8 shows the power output as a function of grid voltage for a 50 cps modulation. VOLTAGE (VOL1s) CONTROL RIMD This characteristic remains unchanged for modulating Fig. 7-Power output and beam current as functions of control grid voltage. frequencies up to at least 5 mc. The control grid is an electrode surrounding the filament. The 'cutoff" is relatively high. It is also posAmplitude Modulation sible to control the current with the plate of the optical In Fig. 6, power has been plotted as a function of system. In this case the modulating voltage is lower. beam current for different frequencies. For these experi- But as shown in Fig. 9 the power vs plate voltage curve ments a tungsten filament was used and the current is sometimes more complicated because the shape of the controlled by the temperature of the filament. trajectories is influenced by the plate voltage. Usually, these curves are straight lines and power outInfluence of Load put is of the form: If attenuation at the end of the line near collector P = A(I-I.), gives a perfect match and there is no reflection along circuit from machining irregularities, frequency will be inI, being the starting current and A a constant.

1955

Warnecke, Gugnard, Doehier, and Epsztein: The "M"-Type Carcinotron Tube

419

I<

Fig. 10-Lines of const ant power on a Smith chart.

dependent of load and curves of constant power must be circles about the center of the Smith chart. Fig. 10 is the Rieke diagram for constant power. Variation of frequency was below measurement accuracy (1 mc). Fig. 11 (next page) shows frequency vs sole-voltage for an imperfect attenuation at end of line near the collector. Parameter is vswr introduced in the output circuit. The form of the curves agrees with theory., Signal to Noise Ratio The measurement of the signal-to-noise ratio is possible, if the "M" carcinotron itself is used as local oscillator. In this case, the variations of frequency due to the ripples in the power supplies cancel out, but it is not possible to separate the two noise sidebands. The noise of the Carcinotron was compared with a mercury-argon

noise source amplified with a low-noise traveling-wave tube. Fig. 12 (next page) shows noise per cps to signal ratio as a function of frequency distance from oscillating frequency. Noise per cps to signal ratio is of the order of - 140 db. Fig. 13 (page 421) shows the noise in relative units as a function of current.

Efficiency
Electronic Efficiency: If space charge is neglected the electronic efficiency fle can be calculated with the same method as in the magnetron amplifier. If all electrons are absorbed by the line, -q, is given by: Vf e t 1 - m-X Vf

V, is the voltage corresponding to the drift velocity of

420
F

PROCEEDINGS OF THE IRE

A pril

Fig. 11-Frequency vs sole voltage for various values of load vswr (ordinate curves have been translated for the various curves).

?iC/S a 1 cps bandwidth. The abscissa is Fig. 12-Noise-to-signal ratio for the difference between the oscillating frequency and the frequency at which noise is measured.

0,01

0,1

so

100

the electrons given by (1). m is a constant depending on attenuation of the line in Neper/cm, we have: rL the trajectories. If the trajectories are linear, K is unity; PL = 2yaf P (y)dy, if they are cycloids corresponding to a plane magnetron without space charge, we have m =4. Actually, not all electrons are absorbed by the line. where P'(y), power propagating along the line, is related Measurements on distribution of collector and line cur- to W(y): rent indicate that more than 80 per cent of electrons arP (y) = W(y)vg rive on the line, and electronic efficiency should be: and therefore: - m ) 77e _0.81 c kg

which can be transformed to: / m /Bclr\\

71e _

Qv=-, ko=-. V 2Q a
a

0.8(1 --(. j),

Qezt is given by:

k fL W(y)dy

Bcr being the cut-off magnetic field of a plane magnetron under the same conditions. Circuit Efficiency: The circuit efficiency is given by the Qv of the circuit and the external Qext
Xc

Qext =

W(0)

Qv Qext + Qv
rL

For small signal theory W(y) can be obtained by the superposition of the waves with the propagation constants given by (2), and we obtain:

Q, is defined by:
W(y)dy
and:
Q =
PL rL

Qext = k5 =YQv(GZ2 'Ya

= =c
-

al

1)

(14)

Ta 1

(15)

and Qext by:

X

If the attenuation is small (14) gives:

W(y)dy

Qext =

l,

(16)

and found directly, if losses in the line are neglected. Influence of the Coupling Impedance on the Efficiency: W is the stored energy per unit length, PL the power According to the theory neglecting space charge, effiloss in the line, and Po the output power. If y. is the ciency should not depend on coupling impedance for

Qz

PO

1955
RELATIVE NOEl

Warnecke, Gugnard, Doehler, and Epsztein: The "M"-Type Carcinotron Tube

I,
a
____ _
0

421

.-e

."

jo KC
X

1-1

a. 1170 GAUSS

~ ~
6[

4.4

201

-a
4-

5000

7500

10000

12500

15000

AX+
it

100

200 BEAM CURRENT (MA)

300

Fig. 14-Efficiency as function of voltage exhibiting rising-sun effect (average over 16 tubes).

Fig. 13-Relative noise

as a

function of beam current.

At the "cutoff" (,6= 7r), this is the well-known condiRising-Sun Effect'3 tion for the "rising-sun effect" in the magnetron'6 charIn the small signal theory, it is generally assumed that acterized by a pronounced minimum in efficiency. As interaction takes place with one space harmonic. How- shown by Fig. 14 above, same phenomenon appears in the "M" Carcinotron for values of parameters satisever there is the possibility that one space harmonic ro to the beam wave re, another space har- fying (19). It is shown in Appendix B that simultanebeing coupled monic r'o will be coupled to the beam waveF -COr/ve. ously the starting current must exhibit a maximum. This can be seen on Fig. 15 (next page). This will occur if:

high beam currents and should be a function only of the shape of trajectories, phase velocity of the circuit, and acceleration voltage. However, the first tubes tested have given a relatively low efficiency, much lower than predicted by theory. It has been assumed space-charge phenomena have an important influence on efficiency (diocotron effect)'2 and if coupling impedance is high, space-charge effects should be negligible and efficiency should increase. It can be seen that, contrary to the space-charge neglecting theory, efficiency increases rapidly with coupling impedance. The same occurs with the tuning range. As far as power is concerned, there is optimum value for the coupling impedance: an increase of coupling impedance decreases the thermal dissipation of the line and consequently the dc power which can be applied.

If (18) is fulfilled, the space harmonic ro" = ro +2 7r/p will be coupled to the beam wave r.+wcve. However, this effect can be neglected, because of the low coupling impedance of the wave ro". This simultaneous coupling of two space harmonics can occur in certain lines, e.g., interdigital lines, where

The first symmetrical space harmonic being normally used, interaction can occur with the first asymmetrical space harmonic which is in fact the fundamental space harmonic. For the symmetric space harmonics, the apparent pitch of the line is half the real pitch p. 41 being the phase angle along this apparent pitch for the first symmetric space harmonic, (18) can be expressed numerically in the following form:
XB
=

asymmetrical and symmetrical space harmonics exist.'4,15

107OOr

(19)

ro ro' =
-

(Cr
V6

(17)

The differences ro - ro' between two successive space harmonics being 2xr/p, where p is the pitch of the line, (17) implies that:
-=
Ve
12 P.

W,r

27r
p

(18)

duce the bulk of the tube, the line has been curved into form. The electromagnet has been replaced by a permanent magnet. Fig. 18 (page 423) shows the performance charts measured on this tube, and Fig. 19 (page 423)
a circular

CHARACTERISTICS OF AN "M" CARCINOTRON Fig. 16 (page 422) shows an experimental setup for studying properties of "M" Carcinotron. Fig. 17 (page 422) shows same type of tube in an industrial form. To re-

exp6rimentale de l'interacondes de charge d'espace au sein d'un faisceau 6lectronique se d6plagant dans des champs 6lectrique et magn6tique crois6s, " Ann. Radiodlect., vol. VII, p. 252; October, 1952. 13 W. E. Willshaw, G. Mourier, and G. Guilbaud, "Effet de resonance 6lectronique dans les tubes A champs 6lectrique et magn6tique crois6s," Compt. Rend. Acad. Sci. (Paris), (in press).
tion
par

Gu6nard and H. Huber, 'Etude

14 R. C. Fletcher, "A broad-band interdigital circuit for use in traveling-wave-type amplifiers," PROC. I.R.E., vol. 40, pp. 951-958; August, 1952. 16 A. Leblond and G. Mourier, "Etude des lignes a barreaux a
p. 184; April, 1954. G. B. Collins, "Microwave Magnetrons," Radiation Lab. Series, McGraw-Hill Book Co., Inc., New York, N. Y., sec. 3.3; 1948.
16

structure p6riodique," Ann. Radioilect., vol. ix,

422

PROCEEDINGS OF THE IRE

A pril

30_

X
4

_- \

--+ s
-

/
4

{ ~

20

__I

ea

ows

lW

14f2M

lKOe

loom0

Fig. 16-Experimental setup for studying a linear "M" Carcinotron.

Fig. 15-Starting current curves exhibiting rising-sun effect.

the variation of power, efficiency and frequency with line voltage. These figures show possibility of obtaining, under practical conditions and for frequencies around 3,000 mc, power outputs of several hundred watts in a frequency range larger than half an octave and with efficiencies of the order of 40 per cent. In some cases, efficiencies in excess of 60 per cent have been measured.8

CONCLUSION This paper shows that if the theory does not cover Fig. 17-Industrial model of an "M" Carcinotron. entirely the behavior of the "M" Carcinotron, in particular the space-charge effects, it nevertheless predicts - at5x - i#wsy = -j?IrOL (20) the main features of this type of tube. The main features of the "M" Carcinotron, as com(21) iw,zx - ay = 7rOK pared to other types of electronically tunable tubes, are: (t = X - rve; K = coth ryo). high efficiency, a fairly linear frequency-voltage characteristic, and a low pushing and pulling figure, to- The coupling between the beam and the line is given by: gether with a wide electronic range. I.E* IE*R,r02 These features make the "M" Carcinotron particr = rO =ja 4P ularly advantageous as a high-power electronically2rr*-OLOL* tunable tube. where R, is the coupling impedance APPENDIX A EZE." R, = Without space charge, the motion of the electrons is 2r 2P determined by the field of the line, the components of An infinitely thin sheet of the beam carrying the dc which can be written: current dlo (which is taken positive) contributes to -* -3I. E* through the following term: E0Pi(ca-rF) = - __ E,ei(wt-r,) = ax dy - jr6dI[6xE=* + SyEy*] = -rr*L*dI[ax - jK8y], with: which gives, Io being the current carried by the beam: sinh ry 4) = f L .~ CJ(wt-r.) sinh ryo (22) j(r - ro)rhL = 2Y [R The equations giving the rf components of the motion are, for a linear trajectory x=v0r, y=yo: where
.

j3Y]

1955
VOLTS

Warnecke, Guenard, Doehler, and Epsztein: The "M"-Type Carcinotron Tube

Fig. 19-Frequency, power output and efficiency as functions of line voltage.

Fig. 18-Performance charts of an aM" Carcinotron: (a) Frequency and power output vs magnetic field and voltage. (b) Efficiency vs magnetic field and voltage.
YM 2
=

i1r srI2RXK

Eo
waves, one

Eqs. (20), (21) and (22) determine five

which is:

rsI

OL

6x =

jK6y

6X

by

Assuming
r o-P I <<Ve

The two other waves correspond to - As compared to the waves r1,2, they are weakly coupled to the field of the line, as seen from (20) and (21), which relate amplitudes of the field and of the electron motion. of The potential 4ei(wt-Fx), from which derives the spacecharge field of such a beam, can be calculated from 0. the rf space-charge p and the surface density -poby. The space-charge density:
=
w7.

t.Fig. 20
P

423

x 1-Am.

and

IYM2 <<

+ Po(

Ax +-eby

is equal to zero. The potentials q5,, inside, and 01, q5 outri,2 are the two principal waves for which tj <<Cr. The side the beam (Fig. 20 above) satisfy Laplace equations propagation constants of these two waves are given by and are thus a linear combination of Ery and c-ry. If the sole and the line are remote enough from the (23) beam (Py1>, r(d-y2)>>1), the potentials are: (r r.)(r ro) yM2
-

Vs

i.e.

rP,2
with

rm

vrd2 + yM2
rd

(24)
ro

Oi = A6rs 02 = Be-ri
=

Cery + Dt-rv.

re + ro
2

re
=

The boundary conditions at the two edges of the beam

give

424

PROCEEDINGS OF THE IRE

A pril
ro

oC

P00Y2

2reo

y2(u-v2) + er

pocSyi

2reo

Er (vl-v )

r.
that is if :

r-.,

r.Wr
Vs

er

r'.ro
Vc

2Pr
-

~P

Assuming JA<<1, 4o takes the simplified form:

fc-A

a02r
Wrt

r- -(KA + Y) (I1-rA).

(25)

21r
p

With space charge, the equations of motion are:

) Ox"- -W j 4srsinhry +4)6c sinh ryo _~ +cor OFa sinhPry + fcI aY + 'k=Q2bL .- Jy Ej' dy Lsmnhy

The amplitudes of these two space harmonics are related by the properties of the line, and for the second wave, the propagation constant will be rI"=r-27r/p and the amplitude k45L. The equations of motion (20) and (21) hold for the first space harmonic. For the second space harmonic, the equations of motion are:
-

with
Vs

t/'26X' - jt'Wr,yl = - jr P'k4OL jitl.raXI- V'23y' = qr'kOLK',

f

(28)
(29)

V.- rn'- P
U0

2y

with:
= t

+ wr.

The equation of coupling between the line and the beam must take into account both space harmonics, that is: Introducing in these equations the value of 4 given by I E* = jIorP46xEx* + OyEv* + 5x'E /* + (25), and assuming: t ae2A =- Pr*4L*10 [(ax - jKOy)

rr

Oy'Ep'*I
.

r-

a!

<< ym,

these equations give

rt*

k(Ox'

(26) Eq. (23) becomes (r - rP)(r - ro) = yM2(l s), t being the value of X-rv., given by the theory which neglects space charge. The propagation constants of the where two principal waves are thus given by: 1 + K'l2 r1 l2
(r
-

ro) (r

)= TM2 I

r*

jK'6y') I~~~

(30)

showing that the starting current is increased by the factor 1/A/1-S. APPENDIX B There are now two waves uncoupled to the beam In Appendix A, it has been supposed that one space ('kL=O) corresponding to =0, t'= r, the four factors harmonic was coupled to the beam. Let us suppose now Ax, Sy, Ox', by' being related by two equations: that the space harmonics are coupled to the beam corrt ax -iKsy + - k(Ox' - jK'y') = 0 responding to propagation constants ro and roP=ro - 2r/p, p being the pitch of the line. Both space harAx' = -jay'. monics will be strongly coupled to the beam if:

Im +YM2 (1 ,Vi d2+

(27)

4K

12

CA5D