VIII.
A.
MAGNETRON
TUBE RESEARCH AND DEVELOPMENT
DEVELOPMENT
Dr. S. T. Martin
A. G. Barrett
1.
Testing and Design of High Power 10. 7-Cm Magnetrons
Our efforts have been directed to the solution of problems concerned in the manuAn essential objective of this project is the
facture of good ceramic-to-metal brazes.
fabrication of a magnetron having all-ceramic insulation, in order that the metal may
The titanium
be made as nearly gas-free as possible by high temperature processing.
hydride process was selected as having the maximum advantage,
at least for experi-
mental work, and for several months gave eminently satisfactory results.
More recently we have found it increasingly difficult to make a good window seal
that will withstand our test pressure of 110 lbs/inch 2 and will test vacuum-tight on the
helium leak detector.
Since the seal is made on the edge of a circular piece of aluminum
oxide 0. 080 inch thick and 2. 500 inches in diameter, nondestructive examination is not
possible.
Although superficial examination always shows good solder flow through the
seal area, very small leaks appear at one or more points along the periphery.
Stripping
the metal away from the ceramic usually shows a poor bond.
In the course of experimentation about sixteen such seals have been made in the
search for the culprit factors.
Occasional test seals appear to be sound and the indica-
tions are that a temperature well in excess of the melting point of the BT (Cu-Ag eutectic) solder used, must be attained.
We have found that it is possible to have titanium hydride which does not behave well;
that is,
wetting of the surface is sporadic and takes place very slowly.
Other factors of
importance are the kind of film of hydride laid on the surface, whether too thick or too
thin; the particle size of the hydride, which affects its quality as a paint; and finally,
the handling of the completed and dry film.
Among the test seals made in this period was a mockup window made entirely from
kovar and welded to a kovar baseplate fitted for attachment to the tube processing
vacuum system.
Although the seal had a small leak, it successfully withstood the fol-
lowing processing schedule:
400 0 C
15. 25 hours
5000
1. 85
5500
4. 80
6000
23.70
6500
42. 35
7000
16.65
104.6U
hours
-34-
(VIII.
At 700°C the weld opened up.
TUBE RESEARCH AND DEVELOPMENT)
After rewelding, the seal withstood the additional pro-
cessing tabulated below before the weld again opened up.
1.00 hours
450"C
5000
14.50
6000
5. 50
6500
1.05
7000
2.90
7500
a
8. 55
33. 50 hours
The window seal opened up during attempts to reweld the cup to the baseplate again.
These results, though not conclusive, indicate that a properly fabricated window
will withstand high temperature processing without serious limitations.
-35-
(VIII.
B.
TUBE RESEARCH AND DEVELOPMENT)
MICROWAVE TUBES
L. D. Smullin
Prof. L. J. Chu
A. D. Berk
1.
a.
A.
H.
J.
R.
W.
A.
M.
C.
Boekelheide
Haus
Houston
Knechtli
C. E. Muehe, Jr.
H. E. Rowe
L. Stark
Noise and Space Charge Waves
Boundary conditions in a multivelocity electron flow
This report presents an analysis of an electron beam with a velocity distribution and
with arbitrary input conditions.
ity spread is assumed.
The general solution can be simplified if a small veloc-
The results are compared with the results obtained from the
single velocity theory.
The assumptions and approximations used in the analysis are a) infinite parallel
plane geometry, i. e. one-dimensional flow; b) continuity of the electric field (short
range collisions are neglected;
c) small signal theory.
The fundamental equation is
2 J (x, v )
l(X v
2
ax
8J (x, v )
+o
1
+ 2
8x
o
with the following notation:
2
Vo)
-
v
p (v)
o
z
m
e
J 1 (x,Vo)
Ev
o
J
o
1 (x, v)
d
(1)
-co
J 1 (x, vo) is the alternating current density per unit velocity;
x is the space coordinate; vo is the time average of the velocity of the stream of elec-
a is the angular frequency; po(Vo) is the direct current
trons in the range v o , vo + dvo;
space charge density per unit velocity; e/m is the charge to mass ratio,
e/m > 0.
The boundary conditions simplify if the current density per unit velocity is a continuous function of the velocity for all x.
Then a Laplace transformation of Eq.
1 yields
the following solution
0
1
(x,
) dv
0
SJ
Jl(0, vo) dV
-o
-
C-joc
f e 1
dv
m
00
*See the Quarterly Progress Report, October 15,
in Eq. 1 should read
.
8v
(c x) 8x
Vo(C,
-36-
(2)
o
(-jvo 0 P)
1951,
p. 31.
2
v(c2
v(c,
x)
ePxdp
o
(c
x)
The third expression
TUBE RESEARCH AND DEVELOPMENT)
(VIII.
where p is the Laplace transform variable.
We express the velocity of a stream in terms of the average velocity of the beam
and its deviation from it:
Equation 2 can be evaluated if the moments of
= w + c.
v
the space charge density and the alternating current density
o()(c)
n
dc
and
f
n
d
0) (WC)\W/
J 1 (O, vo)(
1
- 00
The result is
are neglected for n > 2.
00
Jl(x, vo) dv ° = e - j yx (A+B) cos 6,yx + j
-
B sin 6yx
where
where
00oo
A =
J1 (0, Vo) dvo
2
B=
de - 3
1(0, vo)
-p-
C
- 000
J(0,
J
P-
p
e
m,
v )
1
-00
-
c2
J1(0, vo) dv
-
m
"
w
dv
E
p o(vo)(
dc
o
dc
fPo(v ) dc
w
6= YP
3 c
2
3 c2
2 -77
2 7
w
1+3
y= l a
(W
=
2
.
2
-37-
-
B
(VIII.
TUBE RESEARCH AND DEVELOPMENT)
Under the condition of small velocity spread, three waves are sufficient to describe the
excitation of the beam.
The third wave arises from a continuous wave spectrum which
condenses to one wave in the limit of small velocity spread.
The coefficients A and
A corresponds to a current modula-
C find their analog in the single velocity theory.
tion; C has the role of a velocity modulation; the coefficient B has no parallel in the
single velocity theory.
These results are not altered in content if the assumption of the continuity of the
current density distribution, J1(X, vo),
b.
is dropped.
H. A. Haus
Experimental study of noise in electron beams
The gun previously described (1),
has proved unsuitable, due to focusing difficulties.
A new gun has therefore been installed with the following characteristics.
K = 0. 11 X 10 - 6 amp/volt3/2,
r
k
perveance
rk = 0.070 inch, cathode radius
b = 0. 025 inch, beam radius
= 0. 571 inch, cathode sphere radius
0 = 7. 2,
r = 0. 260 inch, anode sphere radius
half-angle of convergence
at V k = 2000 volts, Ik = 10 ma.
In contrast with the first gun (2), the beam shape of the new gun is very little affected
by thermal velocities.
Because of the higher current density of this gun, the plasma
wavelength is smaller, and more periods of the standing wave pattern can be observed.
The four curves in Fig. VIII-1 are typical of the data obtained with this gun.
these curves the noise power output of the resonant cavity in decibels,
tional to the mean-square noise convection current in the beam,
In
which is propor-
and the percent of the
total beam current which is intercepted before reaching the collector are plotted as
functions of distance along the beam for different magnetic fields.
The absolute level
of the decibel curves is at present arbitrary, but it is the same for all curves,
so the
different curves may be directly compared with each other.
Although these data are in some respects very similar to those obtained for the
previous gun (2), there is at least one important new phenomenon.
At the lowest magnetic field, approximately equal to the Brillouin field, the noise
behaves as shown in the first graph of Fig. VIII-1.
The first and third minima are
approximately equal, while the second minimum is approximately 5 db lower.
As the
magnetic field is raised slightly, this middle minimum rises to the level of the first
and third minima.
Increasing the magnetic field further raises the minima until finally
at very high magnetic fields the noise is approximately independent of distance;
all
traces of the standing wave pattern, with the exception of the first minimum, have disappeared.
This behavior is qualitatively similar to that obtained with the previous gun.
The new, and striking, phenomenon is the appearance of a large increase in noise
beyond the third minimum at low magnetic fields.
-38-
This increase is
not critically
V= 1300V
B=195gouss
I=4.5ma
-7
P=7xIO mm
I,<IupA
VIAS=3 5V
NOISE POWER
I
10
12 14 16 18 20 22 24
D (cm)
26 28 30
32 34
36 38 40
12 14
10
42
16
18 20 22
24 26 28
30 32
34 36
38 40 42
(b)
D (cm)
( a)
24
V 1300V
L =45ma
-P= 7xIomm
22 -
I
10 -
20
9
18
8 -
16
7
14
6
- 12
tO
4
- "8
9-'5
3
2 6
2 -
14
B 715gouss
1,<
IpA
VBIAS 35V
NOISEPOWER
h%
z 2,
U~
1
U
D (cm)
D(cm)
(c)
U148
1
10
12
14
16
822242260334634
18 20 22 24 26
28
30 32
34 36
38 40
(d)
Fig. VIII-1
Noise and partition current vs distance for various magnetic fields. Operating data:
Vk = cathode voltage; I. = total current; P = mm of Hg, pressure; B = magnetic field
in gauss; Brillouin fietd = 200 gauss; percent Ih = percent total current intercepted;
D = distance along beam in cm. Beam enters drift tube at D = 9 cm. See text for
gun characteristics.
(VIII.
TUBE RESEARCH AND DEVELOPMENT)
dependent upon the magnetic field, as is the depth of the middle minimum near the
Brillouin field, but disappears gradually as the magnetic field is increased. It corresponds to an exponential increase in noise with traces of the standing wave pattern superimposed. This behavior may indicate that there is present some amplifying mechanism
which might be used to construct an amplifier.
The partition current, which is also plotted in Fig. VIII-1, is presumably intercepted by the front face of the cavity. At low values of magnetic field it is quite large,
ranging from 4 percent to 7 percent of the total beam current. At high fields it drops
to less than 1 percent. Although we can always increase the measured noise by artificially increasing partition current (by moving the aperture across the beam), we see
that the deepest minimum is the second minimum at approximately the Brillouin field,
where the partition current is quite large.
Since this same general trend has been observed on several different gun designs,
it is probable that the noise is a result of two competing processes; noise increasing
with magnetic field and partition current; but partition current decreasing with magnetic field.
It has been observed that at large distances and small magnetic fields, the region
of operation where the exponential increase of noise is obtained, the intercepted current and the collector current have some type of low frequency instability. It is not
known whether this instability is associated with the exponential gain, although at
certain critical adjustments it is possible to eliminate the beam instability without
altering the noise observed.
In order to obtain more data on these various phenomena, experiments will be performed using two cavities which are tuned to the same frequency. There are at least
two interesting types of experiment: 1) Use the first cavity as a series impedance to
the beam and use the second cavity to measure the noise pattern as a function of the
position of the first cavity. 2) Drive the first cavity with a sinusoidal signal and use
the second cavity to measure the current modulation on the beam. In other words, the
system may be used as a klystron with two movable cavities. In this way experimental
data may be obtained on the propagation properties of the beam at various magnetic
fields. If there is some amplifying mechanism present, as postulated to account for the
exponential increase in noise, it should be possible to construct an amplifier in this
manner.
H. E. Rowe
References
1.
Quarterly Progress Report, Research Laboratory of Electronics, M.I. T. Oct. 15, 1951
2.
Quarterly Progress Report, Research Laboratory of Electronics, M. I. T. July 15, 1951
-40-
TUBE RESEARCH AND DEVELOPMENT)
(VIII.
The demountable system described in previous Quarterly Progress Reports has
The electron beam
been constructed and the first electron gun has been tested in it.
did not lie on the true axis of the system and steps have been taken to correct this fault.
A helix coupling unit which will allow a helix to be placed around the beam and
A movable gun
moved parallel to the beam has been completed and tested for match.
mount has been built which will allow the electron gun to be moved relative to the start
of the magnetic field while the tube is in operation.
gun is mounted,
It consists of a piston in which the
a cylinder in which the piston moves, a rod and bellows to move the
piston, and sliding electrical contacts to supply the necessary power for the gun.
The first experiments on this apparatus will be the measurement of the beam shape.
To accomplish this, a sliding carbon film target will be mounted in the drift tube, movable from the outside.
The incandescent spot will be viewed through a telescope, looking
through an optically flat glass window in the end of the drift tube.
C. E. Muehe, Jr.
2.
a.
Traveling Wave Amplifiers
3-cm pulsed amplifier
The all-glass model described in previous Quarterly Progress Reports was tested
with insufficient loss to prevent oscillations.
At about 9 kv and 0. 4 amp of pulsed cur-
rent in the beam, the tube oscillated with a pulsed power output of about 1 kw.
The
tube has been mechanically redesigned and the first of the new models is almost ready
to be assembled.
L.
b.
D. Smullin
10-cm pulsed traveling wave amplifier
This tube has been designed to explore some of the problems connected with high
power traveling wave tubes.
It will operate with a pulsed beam power of about 100 kw,
and it is hoped that the radiofrequency output will be at least 20 kw.
design is quite conventional:
The mechanical
the helix is directly supported on the glass envelope, and
power is coupled in and out through conventional waveguide transitions.
vided by an Aquadag film on the outside of the glass envelope.
ventional Pierce design,
Loss is pro-
The gun is of the con-
and is scaled from a design by Dr. J.H. Bryant of the Federal
Telecommunications Laboratories.
The initial electrical and mechanical design of the
tube was carried out by Mr. G. C. Dewey of the Weapons Systems Evaluation Group,
Washington, D.
C.,
while he was a guest at the Research Laboratory of Electronics.
The main parameters of the tube are
-41-
(VIII.
TUBE RESEARCH AND DEVELOPMENT)
3000 Mc/sec,
2. 1 cm, helix wavelength
operating frequency
0. 5 cm, mean helix radius
3.0 cm
0. 3 cm, mean beam radius
17,
36 cm, helix length
37 , half-angle of gun
1. 18 turns/cm,(helix pitch)-1
1.8 x 10
15. 1 0, helix pitch angle
540 gauss, axial focusing field.
= 2Tr/k
number of wavelengths along helix
amp/volt3/2
,
gun perveance
The predicted operating conditions are
C= 0.18
I = 5 amp
Q= 2.0
V = 20 kv
Net gain = 40 db.
Cold loss = 40 db
Effect of loss on helix wavelength
Measurements have been made of the effect of the Aquadag coating on the helix
These measurements showed that loss had a pronounced effect on X . This
g
effect is important since it means that the helix wave and the electron beam are not in
wavelength.
synchronism in the lossy section if they are made synchronous in the nonlossy sections
The measurements of X were made by using a movable mercury surface
g
as a short circuit for the helix. Xg was first measured along a plain nichrome helix,
of the helix.
then along the same helix in a glass tube of which part was plain glass and part was
glass covered on the outside with baked-on Aquadag,
and finally along the same tube
and helix with a sheath of metallic foil wrapped around the outside of the glass.
The
dimensions and results were
2. 5 = turns per inch of helix
0. 332 inch = inside diameter of helix
0. 458 inch = outside diameter of helix
0. 478 inch = inside diameter of glass tube
0. 620 inch = outside diameter of glass tube
400 ohms/square = d-c resistance of Aquadag
4. 4 db/inch = r-f attenuation of Aquadag
0. 0003 inch = Aquadag thickness
X
= 10. 00 cm = free space wavelength
X
g
K
g
K
g
K
g
= 3. 68 + 0. 02 cm = for helix alone
o
= 2. 76 + 0. 01 cm = for helix in glass
= 2. 19 + 0. 06 cm = for helix in Aquadag-coated glass
= 2. 34 + 0, 03 cm = for helix in glass covered by nickel foil.
Two possible solutions to the problem of keeping the helix wave velocity equal to
the electron velocity both in and out of the lossy section are 1) vary the helix pitch
-42-
TUBE RESEARCH AND DEVELOPMENT)
(VIII.
between the lossy and lossless sections;
the lossless section of the helix.
2) introduce a metal cylinder coaxially around
Although this will slightly reduce the impedance of
the lossless section, the reduction in gain will be small.
The first tubes will be built with constant pitch helices, and an outer shield will be
used to equalize phase velocities.
J.
c.
M.
Houston
Interleaved-fin slow wave structure
The structure shown in Fig. VIII-2 has been considered as a slow wave circuit for
traveling wave amplifiers.
For traveling wave tube design it is necessary to determine
1) the phase shift between corresponding points separated by one period of the structure;
and 2) the circuit impedance
Z defined by
Z =
2
p1
where E is the amplitude of the accelerating electric field of the synchronous circuit
wave;
p is the propagation constant of the synchronous wave; Pt is the total power pro-
pagating down the circuit.
Both results can be obtained analytically with certain simplifying approximations;
the phase shift calculation can be checked experimentally from measurements of the
resonant wavelengths of a short-circuited length of line.
The approximations used in
the calculations are a) The presence of the holes in the fins is neglected.
b)
The
fields are considered uniform in the X direction within the structure and of negligible
amplitude outside.
This approximation fails completely in certain so-called forbidden
regions which will be defined later.
When L is small compared with free space wavelength and a is comparable with
free space wavelength (in practice this will be the case), the structure can be analyzed
as a folded parallel plate transmission line, loaded at the 1800 bends. The field problem
is thus reduced to the simple analysis of a lump-loaded transmission line.
X
Fig. VIII-2
The interleaved-fin structure.
-43-
(VIII.
TUBE RESEARCH AND DEVELOPMENT)
Fig. VIII-3
Plot of normalized susceptance of the 1800 bend
vs kL/r with d/L as parameter.
The equivalent circuit of the 180
was derived.
0
bend could not be found in the literature,
so it
The exact solution to the problem entails the solution of an infinite set of
simultaneous homogeneous equations.
The equations can be solved very accurately
using the approximate method of Slater in his Technical Report No. 48.
When the ter-
minal planes are located in the surface S, the network proves to be a pure shunt susceptance.
The results of this calculation are plotted in Fig. VIII-3.
Taggart's
measurement of the zero susceptance septum (Radiation Laboratory Report No. 760)
checks very accurately with the curve d/L = 0. 30.
The electric field on the axis of the structure is uniform between fins and delayed
in phase from one section to the next.
With the reference directions for electric field
opposing in adjacent half sections,the phase shift, h L/2, between half sections is given
by
hL
o
cos ---
=
b.'
cos kb - I sin kb
-44-
(1)
TUBE RESEARCH AND DEVELOPMENT)
(VIII.
where
k
o
T0
c
A°
A Fourier analysis of the
b' is the normalized shunt susceptance of the 1800 bend.
electric field on the axis of the structure gives E z as a sum of space harmonics
sin
E
Ez=
h 0Lj(hL
+ n-
e
n2
+ LET
2
4
1
e
-j h o +
n)z
hL
o
o
+n odd
+ n
The dominant space har-
where E 0 is the amplitude of the electric field between fins.
monic at low frequencies is the -1 space harmonic occurring when h L/r
= 1. Choosing
this space harmonic as the synchronous wave the circuit impedance is calculated to be
Z =--
uL
.ul L
tan
kb
kb
hoL sin
cot
2
hL
o
-
)
4
o
4
h L
1T"
(--7---
After the operating point is chosen from this
The impedance is plotted in Fig. VIII-4.
plot, the susceptance of the discontinuity is calculated from Eq. 1.
The phase velocity
is given by
v-_
kL
k
_
2h
-ho
c
-
o
c
1
T h
o
Since v /c is relatively independent of frequency, the last form of the expression for
V_ 1 /c shows that the structure is quite dispersive, especially in the region hoL/2r
%
1.
The phase shift between adjacent sections of the line was checked experimentally
By Floquet's theorem for periodic structures, each
with the setup shown in Fig. VIII-5.
vector field component is given by a sum of terms
F =
z
A ne
n=-oo
If the structure is an integral number of half periods in length, i. e. I = mL/2, then a
complete standing wave can exist when I = q(Xkg/2) = qw/h
the transmission frequencies occur when
h mL
h=q=
-T
0
-45-
0
where q is any integer. Thus
1000
10C
Fig. VIII- 5
Setup used to measure the transmission frequencies.
2
IC
D5
b = iO
605 INCHES
a =0865
L =0.500
W= 2.860
= 47"
0.50
0.40
SUSCEPTANCE
b-
S-.5
Zo0
(-
k
I
1.0
I
1.1
120r OHMS
i -
-
'
-
0.20
CALCULATED
0.10
MEASURED
I
I
1.2
030
= c0
2Xo
h oL IS PHASE SHIFT
OVER ONE PERIOD
kL/2r
hoL/2r-I
-HL
-
1.3
2L
27
I
1.4
I I
1.5
Fig. VIII-4
The impedance of the -1 space harmonic
as a function of h L/2Tr with kb/rr as parameter.
04
hL
21r
Fig. VIII-6
Plot of kL/2r vs hoL/2Tr.
TUBE RESEARCH AND DEVELOPMENT)
(VIII.
or
hL
o
q
T
m
q need only be determined at one of the transmission frequencies; successive transIf the structure is many
mission frequencies are separated by integral values of q.
periods long, then the transmission frequencies are closely spaced.
A typical plot is
shown in Fig. VIII-6.
One can show that there are so-called "forbidden regions" of the propagation constant for periodic open boundary structures (S.
Engineering, M. I. T.
Sensiper:
doctoral thesis in Electrical
In this particular case the forbidden regions are defined
1950).
by
or
7
+1,
(n= 0,
hL
o
+n
<
+...)
kL
-.
The first form shows that the forbidden regions of h
are those in which any space har-
monic has phase velocity greater than the velocity of light in the medium.
forbidden region boundaries are indicated in Fig. VIII-6.
A few of the
As ho approaches the first
forbidden region boundary, the phase velocity of the -1
space harmonic approaches that
When this is the case, the amplitude of the -1
space harmonic falls off slowly
of light.
in the transverse plane.
The assumption (b) used in the calculations is then a poor one
and the impedance will be somewhat less than the calculated value.
L.
3.
Stark
Spiral Beam Oscillator
The tube and magnet have been redesigned to get a denser electron beam and a more
uniform and intense axial magnetic field.
Several simplifying sets of assumptions have been considered in order to make a
theoretical analysis possible.
There is,
at present, nothing definite to report in this
direction.
A. D. Berk, R.
4.
C.
Knechtli
l-Mev Pulsed Electron Source
The high-voltage,
cathode ray tube was reassembled with a new glass envelope,
which proved to be unsuccessful.
Because of this and the previous failure,
design (1, 2) of the glass-to-metal seal has been changed.
the
The 8. 05-inch diameter
glass envelope has been lengthened so that the seal is located behind the accelerating
anode, and the diameter of the seal has been reduced from 8. 05 inches to 5. 75 inches.
-47-
(VIII.
TUBE RESEARCH AND DEVELOPMENT)
Another tube, incorporating this new design, is in the process of assembly.
A. W. Boekelheide
References
1.
Quarterly Progress Report, Research Laboratory of Electronics, M. I. T. Jan. 15,
1951
2.
Quarterly Progress Report, Research Laboratory of Electronics, M. I. T. Oct. 15,
1951
-48-