High-power operation of a 170 GHz megawatt gyrotron

K. E. Kreischer, T. Kimura, B. G. Danly, and R. J. Temkin

Citation: Phys. Plasmas 4, 1907 (1997); doi: 10.1063/1.872334

View online: http://dx.doi.org/10.1063/1.872334
View Table of Contents: http://pop.aip.org/resource/1/PHPAEN/v4/i5
Published by the AIP Publishing LLC.

Additional information on Phys. Plasmas

Journal Homepage: http://pop.aip.org/
Journal Information: http://pop.aip.org/about/about_the_journal
Top downloads: http://pop.aip.org/features/most_downloaded
Information for Authors: http://pop.aip.org/authors

Downloaded 28 Jul 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
 High-power operation of a 170 GHz megawatt gyrotron*
K. E. Kreischer,† T. Kimura, B. G. Danly,a) and R. J. Temkin
Plasma Fusion Center, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
~Received 11 November 1996; accepted 15 January 1997!
Recent gyrotron oscillator experiments have achieved record powers at 170 GHz. Single mode
emission with a peak output power of 1.5 MW and an efficiency of 35% has been measured. The
experiment is based on a resonant TE28,8,1 cylindrical cavity situated in a 6.7 T magnetic field.
Microwaves are generated in the cavity by an 83 kV annular electron beam produced by a
triode-type magnetron injection gun that is capable of currents up to 50 A. Megawatt power levels
with efficiencies between 30%–36% have been measured over a wide range of operating parameters
for the TE28,8,1 mode. Similar results were also achieved in the neighboring TE27,8,1 mode at 166.6
GHz, and the TE29,8,1 mode at 173.5 GHz. The high output power is the result of a carefully
designed electron gun with low perpendicular velocity spread ~6%–10%! and a novel cavity with an
output iris that is less prone to mode competition. These results are in good agreement with
nonlinear multimode simulations. © 1997 American Institute of Physics.
@S1070-664X~97!92205-9#

I. INTRODUCTION walls becomes a limiting factor when high powers are gen-
1 erated.
The gyrotron was first proposed as a high-frequency A variety of potential applications have motivated the
device in the 1950s, and has since developed into a powerful development of gyrotron oscillators and amplifiers,8,9 includ-
source of microwaves. The high average power capability of ing high-resolution radar, particle acceleration,10 material
gyrotrons is particularly evident in the millimeter and sub- processing, and plasma diagnostics.11 Certainly one of the
millimeter wave region when compared to conventional strongest motivations has been the need for powerful milli-
tubes such as klystrons and traveling wave tubes ~TWTs!. A meter wave sources for electron cyclotron resonance heating
large variety of experiments over the past ten years have ~ECRH! of fusion plasmas. Recent experiments on both to-
demonstrated that gyrotrons are capable of megawatt power kamaks and stellarators12 have demonstrated that efficient
levels at frequencies of 100 GHz and above. In the U.S., heating with local deposition of the power is possible with
researchers at C.P.I.2 ~Communication and Power Industries, ECRH. Plasma heating requires efficient gyrotron oscillators
formerly Varian Associates! working at 110 GHz with a capable of producing high average power in a single, high-
TE22,6,1 mode cavity have achieved output powers of 680, frequency mode.
530, and 250 kW for pulse durations of 0.5, 2.0, and 10.0 s, The next major fusion experiment now being planned is
respectively. There is also a significant international gyrotron the collaboration known as the International Thermonuclear
research effort, primarily motivated by the need for long Experimental Reactor ~ITER!.13 This device is expected to
pulse oscillators for fusion plasma heating. Notable oscillator produce over 1000 MW of fusion power, and auxiliary heat-
experiments include efforts at 84,3 110,4 118,5 and 1406 GHz ing will be essential for accessing the H-mode operating re-
that have produced powers between 0.5–1.0 MW for pulse gion and reaching ignition. ECRH utilizing gyrotrons is a
lengths of 0.2–5.0 s. strong contender for this role,14 and is being considered for a
The gyrotron is a fast-wave device that relies on a cy- variety of functions including plasma breakdown, bulk heat-
clotron resonance interaction between an electron beam and ing, and current drive. Frequencies between 140 and 200
rf fields in a TE-mode cavity to generate microwaves. This GHz will be needed depending on the function. Because of
synchronism is maintained by an axial magnetic field, and as the large size of the planned ECRH system ~up to 100 MW!,
a result, a simple cylindrical resonator capable of high power individual gyrotrons must produce at least 1 MW. System
can be used. This is in contrast to conventional slow-wave costs will be important, so each unit will need to achieve an
devices, such as the TWT, that are prone to overheating and interaction efficiency greater than 35%.
breakdown at high frequencies because of the small coupling To circumvent the wall thermal problems associated
structures that are required. The gyrotron operates with the with operating near cutoff, gyrotrons typically utilize highly
cavity near cutoff, and as a result its interaction efficiency is overmoded cavities. To determine a suitable mode for a
less sensitive to beam velocity spread than other fast-wave given power P and frequency n, one must determine the
devices such as the free-electron laser ~FEL!.7 However, be- energy balance within the cavity, and from this calculate the
cause the gyrotron operates near cutoff, the rf fields in the equilibrium rf fields. This leads to the following equation15
cavity are very intense, and thermal heating of the cavity for the mode index n mp of the TEmp mode of a gyrotron
cavity:
*Paper 4IB4, Bull. Am. Phys. Soc. 41, 1474 ~1996!.
†
Invited speaker.
a! 1.231023 L P ~ MW! n 2.5~ GHz!
Present address: Naval Research Laboratory, Code 6843, Washington, ~ n 2mp2m 2 ! 5 . ~1!
D.C., 20375. l ~ 12R 2 ! r Ohm~ kW/cm2!

Phys. Plasmas 4 (5), May 1997 1070-664X/97/4(5)/1907/8/$10.00 © 1997 American Institute of Physics 1907

Downloaded 28 Jul 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
 modes would be needed. This has led to gyrotrons based on
complex and coaxial cavities.
The first experiments at MIT16,17 investigated the
TE0,3,1 and TE15,2,1 modes at 140 GHz. Both gyrotrons were
able to generate high powers, but initially had efficiencies
below 30% at higher beam currents. Careful design of the
cavities, and optimization of the operating parameters, even-
tually eliminated this problem in both cases. We later inves-
tigated very high-order TE modes with n mp up to 75 in the
200–300 GHz range.18 Efficiencies below 25% were mea-
sured at powers approaching 1 MW. We were unable to sig-
nificantly improve this efficiency, and attributed our difficul-
ties to poor beam quality and a tapered cavity with too many
competing modes.
This paper describes the efforts at MIT to demonstrate
efficient gyrotron operation at 170 GHz with a tapered cav-
ity. It is organized in the following manner. In Sec. II we will
give an overview of the design of the experiment, focusing
FIG. 1. The dependence of the TE mode index n mp for a 170 GHz gyrotron on our efforts to build an electron gun with very good beam
on the output power and the peak Ohmic losses in the cavity walls. quality. In Sec. III, we will present our measurements of the
operation of the tube, including efficiency, observed modes,
and start-up characteristics. We will also compare these re-
sults with nonlinear theory. Conclusions and future plans
In this equation L is the characteristic axial length of the rf will be discussed in Sec. IV.
field profile in the cavity, l is the wavelength, R 2 is the
II. DESIGN OF EXPERIMENT
reflection coefficient at the cavity output, and r Ohm is the
peak Ohmic heating density in the cavity walls. An electrical This experiment was our first attempt to build a proto-
conductivity of 5.03107 (V2m) 21 , corresponding to ideal type of a 170 GHz megawatt gyrotron that would meet the
conditions of copper at room temperature, is assumed. In needs of ITER. Therefore we decided to choose conservative
actual operation, the gyrotron cavity wall temperature can be design features that had been successfully demonstrated in
as high as 250 °C, and the resulting higher resistivity can past tests. The experiment was based on a cylindrical tapered
increase the Ohmic losses by about 50%–70%. Equation ~1! cavity that has proven very successful in previous high-
shows the tradeoff between power and frequency, and the power gyrotrons. Even though there was some concern about
strong dependence of the operating mode on frequency. This the cavity’s effectiveness when operating in a very high-
equation has been plotted in Fig. 1 using L/l of 6.5, which order mode, we were able to mitigate this problem by keep-
results in an efficient interaction between the beam and rf ing the interaction length short. A schematic of the 170 GHz
field, and a reflection coefficient R 2 of 0.5. It is clear from gyrotron is shown in Fig. 2. The experiment was operated
Fig. 1 that utilizing advanced cooling techniques that allow with the tube in the horizontal position. A superconducting
the highest values of r Ohm would be advantageous. However, 7.5 T magnet built by Cryomagnetics, Inc., with a 20 cm
tube reliability is also an issue, and excessive cavity wall warm bore provided the cavity magnetic field. The magnet
temperatures can lead to problems such as cyclical fatigue. consists of a pair of coils that allow tapering of the magnetic
Past experience with long pulse gyrotrons have shown that field in the cavity region. Such magnetic tapering has been
peak ohmic losses of about 2 kW/cm2 ~ideal conditions! can used in the past to enhance the efficiency. However, the re-
be tolerated without degrading reliability. For 1 MW output sults presented in this paper were obtained with a flat field at
powers at 170 GHz, Fig. 1 indicates that a mode with n mp the cavity. The large bore is needed to accommodate the
greater than 50 is required. mode converter and mirror transmission line inside the gyro-
At the Massachusetts Institute of Technology ~MIT! we tron. The magnet can be operated up to 9 T by pumping on
are presently investigating gyrotron oscillators that will sat- the liquid helium cryostat. This feature will allow us to in-
isfy the needs of the ITER experiment. Our gyrotrons are vestigate gyrotron operation above 200 GHz. A 0.3 T room-
typically operated for 3 ms pulses at 2 Hz but are designed to temperature gun coil attached to the magnet was used to
model long pulse or cw devices. This allows us to study the adjust the electron beam radius and velocity ratio.
physics of the interaction without the complications associ- The overall design of the 170 GHz gyrotron is similar to
ated with high average power. We are particularly interested that used in many previous experiments, including those at
in characterizing operation in very high-order TE modes, and MIT. Parameters of this tube are given in Table I, and are
determining if mode competition adversely affects the inter- based on peak cavity Ohmic losses of 2.1 kW/cm2 ~ideal
action efficiency as the gyrotron becomes more overmoded. conditions, see Fig. 1!. The major differences of this experi-
Past experiments have suggested that the tapered cavity com- ment when compared to our previous research at 110 GHz19
monly used in gyrotrons becomes less effective when are the need to operate in a higher-order mode, and the larger
n mp>40, and that a more complicated structure with fewer magnetic compression of the electron beam ~36 versus 24!.

1908 Phys. Plasmas, Vol. 4, No. 5, May 1997 Kreischer et al.

Downloaded 28 Jul 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
 FIG. 2. Schematic of the MIT 170 GHz gyrotron.

The annular beam is produced by a triode magnetron injec- beam would couple to the inner radial maximum of the
tion gun built by C.P.I. A beam with voltages up to 100 kV, TE28,8,1 mode. This corresponds to k' R e 529.45, where k' is
and currents up to 50 A, can be produced. The drift region the perpendicular wave number in the cavity. Positioning the
between the gun and cavity contains slotted structures to beam on the inner radial maximum provides stronger cou-
prevent spurious oscillations. In addition, there is a 5 cm pling and less mode competition. The 0.86 cm separation
long section just before the cavity consisting of alternating between the beam and cavity wall is quite large, and results
copper and ceramic rings that absorb any power leaking from in a relatively high voltage depression of the beam ~6 kV for
the cavity that could reach the cathode region and disrupt a 35 A beam!.
gun operation. After magnetically compressing the beam in A major focus of this experiment was to design and con-
the drift region, its velocity ratio a 5 v' / v i ~v' and v i are the struct a highly overmoded tapered cavity less affected by
beam’s perpendicular and parallel velocity! at the cavity is mode competition. In order to determine when competition
1.6. The beam radius R e in the cavity was chosen so that the would prevent operation at high efficiency, the cavity de-
signs were simulated with a self-consistent, time-dependent
multimode code developed at the University of Maryland.20
TABLE I. MIT 170 GHz gyrotron parameters.
We started with a simple profile consisting of a linear up-
Mode TE28,8,1 taper, straight section, and linear output taper. We length-
Voltage ~kV! 83 ened the straight section in order to increase the diffractive
Current ~A! 36 Q to the desired value, but found that multimoding became a
Velocity ratio 1.6 problem when L/l exceeded 6.5. We therefore chose a rela-
Efficiency ~%! 35
Magnetic field ~T! 6.7
tively short straight section to ensure that L/l did not exceed
Field compression 36 this value, but included an iris at the cavity output to raise
Cavity radius ~cm! 1.69 the cavity diffractive Q to the desired value. The final design
Beam radius ~cm! 0.83 of the resonator consisted of the following dimensions: 3°
L/l 6.4 linear input taper, 0.8 cm long straight section with a
Diffractive Q 1300
Peak cavity loss ~kW/ cm2! 2.1
531023 cm iris step at the end, and a nonlinear uptaper to
Voltage depression ~kV! 6.0 7% above the cutoff radius of the cavity. This cavity has a
diffractive Q of 1300 based on cold cavity simulations,

Phys. Plasmas, Vol. 4, No. 5, May 1997 Kreischer et al. 1909

Downloaded 28 Jul 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
 FIG. 4. The dependence of the beam velocity ratio on the magnetic field at
FIG. 3. Intensity of the surface currents on the inside wall of the launcher. the cathode for three different designs.
The contours represent the ratio of the local wall current to the current at the
beginning of the launcher. The contours are plotted as a function of the axial
distance z and the azimuthal angle f.

tron beam. One concern with operation near cutoff was the
possibility that the gyrotron interaction would continue be-
which is about 2.5 times larger than the minimum Q of yond the cavity and into the launcher. However, nonlinear
4 p (L/l!2. Operating well above the minimum Q is needed simulations indicate that the interaction does end early in the
to achieve high efficiency, and helps to reduce the cavity’s nonlinear uptaper of the cavity, primarily due to the drop-off
sensitivity to external reflections. The transition from the in- in the magnetic field. Another concern was the possibility of
put taper to the straight section was rounded in order to mini- spurious modes being excited in the launcher. However, we
mize mode conversion that could cause power leakage back confirmed in our 110 GHz experiments that it is possible to
into the gun region. Mode conversion in the nonlinear up- design such a launcher so that it does not support high-Q
taper was calculated to be less than 0.5%. Assuming a 10% modes, and is therefore less susceptible to such oscillations.
perpendicular velocity spread and a beam velocity ratio of The four mirror transmission line following the launcher
1.6, an efficiency of 39% was predicted by multimode non- is used to phase correct and focus the beam, and then launch
linear theory for an 83 kV, 36 A beam, giving 1.2 MW of rf the radiation through a side window. The transmission line
power generated in the cavity. Ohmic and transmission also separates the radiation from the spent electron beam in
losses inside the gyrotron and at the window reduce the the gyrotron, simplifying the design of the collector. This
power leaving the gyrotron to 1 MW, for an overall effi- arrangement reduces the deleterious effects of reflections
ciency of 35%. from the vacuum window, and provides better pumping con-
The high-order TE modes generated by megawatt gy- ductance within the vacuum tube for quicker processing. Dif-
rotrons are difficult to transmit efficiently, and therefore an fraction codes developed at MIT21 were used to design the
internal mode converter can typically be found in these de- internal mode converter, and an overall conversion efficiency
vices. In our case the mode converter immediately follows of 93%–95% was achieved. The final output is a circular
the cavity, and is designed to transform the TE28,8 mode into Gaussian-like beam with a 2.8 cm waist at the 10 cm diam
a Gaussian beam that can then be transported to the fusion window.
plasma with corrugated waveguide. The converter21 consists An important goal of this experiment was to optimize
of a cylindrical launcher with small periodic wall perturba- the electron gun design so that a high-quality beam was pro-
tions and a four-mirror transmission line. The perturbations duced over a wide range of operating parameters. Our 110
produce a mix of satellite modes that combine with the origi- GHz gyrotron experiments19 indicated that good beam qual-
nal mode to give a Gaussian-like beam. This beam is then ity is critical in achieving higher efficiency. Factors that can
launched through a slot in the launcher wall. Figure 3 shows degrade beam quality include poor beam optics, cathode sur-
the evolution of this mode mix in the launcher. Starting with face roughness,22 electrode machining errors and operation at
only the TE28,8 mode at z50, the satellite modes are gradu- high temperature. The 170 GHz gun was simulated with an
ally produced as z increases, resulting in a sequence of electron trajectory code having a mesh size sufficiently small
peaked current contours corresponding to an rf beam bounc- that the beam optics were accurately modeled. The electrode
ing off the interior wall along a helical path. The box in Fig. shapes were optimized so that the beam characteristics were
3 represents the slot through which the rf beam is launched. insensitive to machining errors and to variations in the oper-
Our launcher was designed to operate close to cutoff in ating parameters resulting from misalignment or beam volt-
order to reduce its length and provide clearance for the elec- age fluctuations. Figure 4 shows one example of this optimi-

1910 Phys. Plasmas, Vol. 4, No. 5, May 1997 Kreischer et al.

Downloaded 28 Jul 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
 zation process. In this graph the dependence of the beam
velocity ratio a on the cathode magnetic field is shown for
three gun designs. All three designs could achieve the de-
sired a value of 1.6–1.7 with low-velocity spread. However,
the design with a less sensitive to the cathode field was
chosen. For this design a slight variation of magnetic field
azimuthally around the emitter would result in a smaller ~Da/
a!, or spread in the velocity ratio. This implies a smaller
perpendicular velocity spread ~D v' / v' ! because, for a mo-
noenergetic beam, the two spreads are related by the equa-
tion Da / a 5(11 a 2 )(D v' / v' !. Also shown in Fig. 4 is the
predicted beam velocity ratio based on an adiabatic theory.23
Clearly adiabatic theory inadequately explains the depen-
dence of a on the cathode field, especially below 1.85 kG.
Simulations of the final design of our 170 GHz gun in-
dicated that the perpendicular velocity spread due to beam
optics is 2.5% for a 35 A beam. Taking into account addi-
tional spreads, which includes 3.7% from the cathode surface FIG. 5. The measured output power, efficiency, and detuning parameter for
roughness, 0.9% from misalignment, and 2.6% from the ma- the TE27,8,1 mode after optimization of the cavity magnetic field.
chining errors, the final beam spread was predicted to be
5%–7%. Nonlinear simulations of the cavity based on this
spread indicated that an operating efficiency of 35% would fying that the beam was propagating through the center of
be feasible. the cavity.
We then investigated the operation of the gyrotron
III. EXPERIMENTAL RESULTS
around the design frequency of 170 GHz. We found that the
TE27,8,1 at 166.6 GHz was easier to access than the TE28,8,1,
Although the overall goal of the MIT program is to in- and focused our attention on that mode. Figure 5 shows the
vestigate the gyrotron configuration as shown in Fig. 2, the dependence of the output power and efficiency on the beam
initial experiments were conducted without the internal mode current. The power was measured with a 20 cm Scientech
converter. The results of these tests are presented in this calorimeter that was modified to be 85%–90% absorbing at
section. The objectives of these experiments were to investi- these frequencies. At each beam current in Fig. 5 the cavity
gate the electron-beam characteristics and to determine if and cathode magnetic fields were optimized to maximize
mode competition was adversely affecting the operation of power. The shot-to-shot power stability was better than 10%
the cavity. It was felt that by operating without an internal at each operating point, and measurements using our har-
mode converter, the measurements would be easier to inter- monic mixer indicated single mode emission at all currents.
pret. The rf power generated by the cavity was propagated by Efficiencies over 30% were achieved for all currents over
a 2.2 cm radius cylindrical copper waveguide along the tube 15 A. Also shown is the optimized detuning parameter
axis to a quartz vacuum window. The launching of the D52~ 12 v c / v )/ b'2 , where v c 5eB 0 / g mc is the cyclotron
TE28,8 mode through the output window also allowed us to frequency and is calculated from the cavity magnetic field
assess the performance of the mode converter externally to B 0 and the relativistic factor g, v is the rf frequency, and
determine if it was functioning correctly. b' 5 v' /c. The increase in D as the beam current rises is
These experiments were carried out with the gyrotron consistent with nonlinear theory. The measured detuning
operating at 2 Hz with 3 ms pulses. We began with low- value of 0.43 at 35 A is lower than the optimum theoretical
voltage operation, and verified that full beam transmission value of 0.5 and may indicate that mode competition is pre-
was being achieved. We then used beam interception on the venting operation at magnetic fields that yield the highest
beam scraper just before the cavity to adjust the tube in the efficiencies.
magnet bore and center the beam in the cavity. Once we Although the TE28,8,1 mode was a little more difficult to
confirmed that the beam was centered and not scraping on excite than the TE27,8,1, we were able to achieve similar high
any internal components, we operated the gyrotron at full powers in both modes. A comparison of operation at high
voltage ~80–100 kV! and swept the cathode and cavity mag- current in these two modes can be seen in Fig. 6 for beam
netic fields to obtain a spectrum of modes. Thirteen modes voltages between 83 and 87 kV. The results were nearly
were identified between 166 and 178 GHz, an indication of identical, which was expected because the coupling between
the dense spectrum of the cavity. The frequencies of these the beam and rf field is similar for both modes. A maximum
modes were measured to within 5 MHz using a harmonic output power of 1.5 MW was measured in the TE28,8,1 mode
mixer and digital scope with fast Fourier transform capabili- at 170.0 GHz with an 87 kV, 49 A beam for an efficiency of
ties. These frequencies agreed with predictions based on cold 35%. The velocity ratio of the electron beam was measured
cavity theory to within 650 MHz. This confirmed that the to be 1.4. When operating at the design point listed in Table
cavity had been built to the correct dimensions. The TEmp1 I, 1.05 MW of output power was generated with an effi-
modes observed had radial indices p between 7 and 10, veri- ciency of 35%.

Phys. Plasmas, Vol. 4, No. 5, May 1997 Kreischer et al. 1911

Downloaded 28 Jul 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
 FIG. 6. A comparison of the operating characteristics of the modes
TE27,8,1 ~squares! and TE28,8,1 ~circles! at high current.
FIG. 7. A map of the observed modes as a function of the cathode and
cavity magnetic fields.

An interesting comparison can be made between the

TE27,8,1 operating points near 40 A in Figs. 5 and 6. This
of 40, indicates the region where electrons start to reflect
comparison is given in Table II. Both operating points, al-
back into the gun because their velocity ratio is too high.
though quite different, produce 1.25 MW and an efficiency
Between these two boundaries we primarily observed two
of about 35%. Although one might expect that operation at a
rows of azimuthal TEmp1 modes with radial indices p of 8
higher velocity ratio ~column 1! would yield a higher effi-
and 9. Each row corresponds to a region of constant com-
ciency, the lower detuning ~0.44! indicates that mode com-
pression with a fixed beam radius R e . For the upper row, the
petition is limiting access to the high-efficiency region. The
beam is larger and couples to modes closer to the cavity wall
operating point with lower a ~column 2! has less energy
~i.e., modes with lower p!. The observed sequences of azi-
available for microwave generation, but the higher detuning
muthal modes with the same radial index is consistent with
suggests this energy is being extracted more efficiently.
the fact that the gyrotron coupling factor J m61 (k' R e ! is a
In order to better understand the severity of mode com-
weak function of m when m@1.
petition and multimoding in our experiment, a map of the
In general, our gyrotron was characterized by regions of
observed modes was produced. Oscillations in the competing
single mode emission. Figure 7 indicates a few operating
mode can result in a decrease of the interaction efficiency,
regions that resulted in multimoding, such as the boundary
degraded performance of the mode converter, and an in-
between the TE24,8,1 and TE25,8,1 modes. It is also evident
crease in window reflections that can cause trapped power in
that the TE28,8,1 region of excitation is smaller than the re-
the tube. This mode map was produced by varying the cath-
gions of neighboring modes. This may be due to the fact that
ode and cavity magnetic fields, and is shown in Fig. 7. Vary-
the window was matched to the TE28,8,1 frequency of 170
ing the cathode field changes the beam velocity ratio, with
GHz. For other modes at different frequencies, the window
lower field corresponding to higher a. Varying the cavity
mismatch can cause reflection and raise the cavity Q, in-
field allows one to optimize the detuning parameter D. The
creasing the potential for exciting these modes. One indica-
upper boundary of the region where modes are excited cor-
tion that window reflections may be influencing the gy-
responds to a magnetic compression of about 35. Above this
rotron’s operation is shown in Fig. 8. This graph shows the
boundary the compression is less, resulting in a larger beam
measured frequency while the detuning parameter was var-
that partially intercepts the circuit between the gun and cav-
ied. The discrete jump of 70 MHz observed for the TE27,8,1
ity. The other boundary, which corresponds to a compression
mode is consistent with a window reflection.24 For the
TE28,8,1 mode the frequency varies continuously, indicating
TABLE II. TE27,8,1 operation at high power. no such reflection.
We also investigated the influence of start-up on the op-
Operating point 1 2 erating behavior. It has been shown25 that the way in which
Voltage ~kV! 94 84 the operating point is reached can be used to eliminate un-
Current ~A! 38.7 42.8 desirable modes and assure oscillation in the proper mode.
Power ~MW! 1.25 1.25 Figure 9 shows the start-up characteristics of the TE27,8,1
Velocity ratio 1.94 1.43 mode as the cathode voltage was raised to 105 kV. The
Efficiency ~%! 34.6 34.4
Magnetic field ~T! 6.64 6.55
nonadiabatic behavior of the gun is clearly evident, with the
Detuning 0.44 0.59 oscillatory dependence of a on cathode voltage. Also shown
are the theoretical starting current curves for the modes with

1912 Phys. Plasmas, Vol. 4, No. 5, May 1997 Kreischer et al.

Downloaded 28 Jul 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
 FIG. 8. The change in output frequency as the cavity magnetic field was
varied. The dotted line represents a discrete jump in frequency.

FIG. 10. A comparison of the beam velocity ratio as measured with a

the strongest coupling to the beam. No microwaves were capacitive probe with predictions based on electron gun simulations.
observed up to 63 kV. At that point, the TE29,8,1 was excited,
consistent with the theoretical starting current. The excitation
of higher-frequency modes before the TE27,8,1 is reached is probe located directly in front of the cavity. This graph ex-
consistent with the fact that the cyclotron frequency is higher hibits the same nonadiabatic behavior as was seen in Fig. 9.
at lower beam voltages. Note that each mode continued to The experimental points deviate slightly from the theoretical
oscillate beyond the region predicted by linear theory, an predictions, but overall the agreement is good.
indication that the mode was operating in the hard excitation
region.
The high power and efficiencies measured in this experi- IV. CONCLUSIONS
ment indicate that the electron gun is producing a good qual- The results of initial experiments on the MIT 170 GHz
ity beam with the expected characteristics. A more direct megawatt gyrotron have been presented and indicate that the
confirmation that the gun is operating as designed can be overall performance of the tube was excellent and consistent
seen in Fig. 10. This graph shows the dependence of the with theoretical predictions. Megawatt power levels with ef-
velocity ratio on the cathode voltage. Measurements are ficiencies between 30%–35% were measured over a wide
compared with predictions based on electron trajectory simu- range of operating parameters for the TE28,8,1 mode. Similar
lations. The velocity ratio was measured using a capacitive measurements were also obtained in the neighboring
TE27,8,1mode at 166.6 GHz, and the TE29,8,1 mode at 173.5
GHz. These results demonstrate that efficiencies above 35%
can be achieved in gyrotron cavities with n mp.50 and rep-
resent a major achievement. Particularly encouraging was
the fact that the high efficiencies could be maintained at the
highest output powers, in contrast to some previous experi-
ments where the efficiency tended to degrade at the higher
beam currents. These results confirm that the careful design
process has produced an electron gun with a high-quality
beam, and that the tapered cavity with an output iris is effec-
tive in reducing mode competition.
In the next phase of this experiment, the mode converter
will be tested externally to confirm that it can efficiently
produce a Gaussian beam. It will then be placed inside the
gyrotron so that we can evaluate the performance of the com-
bined cavity and mode converter. In later experiments we
will explore the option of using the same gyrotron to produce
discrete frequencies at 150, 170, and 190 GHz by varying the
cavity magnetic field. These frequencies were chosen be-
cause they would be matched to the vacuum window. Simu-
lations indicate that good beam quality, and cavity interac-
FIG. 9. A comparison of the modes observed during start-up with predic-
tions based on linear theory. The scan was done for a magnetic field of 6.77 tion efficiencies of 35%, should be possible over this
T. The beam current varied from 21.2 A at 60 kV to 26.9 A at 104 kV. frequency range. We will also investigate the use of single-

Phys. Plasmas, Vol. 4, No. 5, May 1997 Kreischer et al. 1913

Downloaded 28 Jul 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
 stage depressed collector that could increase the overall effi- 7
C. W. Roberson and P. Sprangle, Phys. Fluids B 1, 3 ~1989!.
8
ciency to 50%. As part of this study, we will study the role of E. V. Zasypkin, M. A. Moiseev, I. G. Gachev, and I. I. Antakov, IEEE
Trans. Plasma Sci. PS-24, 666 ~1996!.
trapped electrons and their effect on the performance of the 9
Q. S. Wang, D. B. McDermott, and N. C. Luhmann, Jr., IEEE Trans.
gyrotron. Plasma Sci. PS-24, 700 ~1996!.
10
V. L. Granatstein and W. Lawson, IEEE Trans. Plasma Sci. PS-24, 648
ACKNOWLEDGMENTS ~1996!.
11
J. S. Machuzak, P. P. Woskov, J. Gilmore, N. L. Bretz, H. K. Park, R. E.
The authors wish to thank W. J. Mulligan and G. F. Aamodt, P. Y. Cheung, D. A. Russell, and H. Bindslev, Rev. Sci. Instrum.
Yarworth for their help in the assembly of this experiment. 66, 484 ~1995!.
12
They also thank Kevin Felch and his colleagues at C.P.I. for V. Erckmann and U. Gasparino, Plasma Phys. Controlled Fusion 36, 1869
their involvement in the construction of the electron gun, and ~1994!.
13
P. H. Rebut, Fusion Eng. Design 27, 3 ~1995!.
Baruch Levush and Tom Antonsen at the University of 14
M. Makowski, IEEE Trans. Plasma Sci. PS-24, 1023 ~1996!.
Maryland for their collaboration on the design of the cavity 15
K. E. Kreischer, B. G. Danly, J. B. Schutkeker, and R. J. Temkin, IEEE
used in these experiments. Trans. Plasma Sci. PS-13, 368 ~1985!.
16
This work was supported by the U.S. Department of En- K. Kreischer, J. Schutkeker, B. Danly, W. Mulligan, and R. Temkin, Int. J.
ergy under Contract No. DE-FC02-93ER54186. Electron. 57, 835 ~1984!.
17
K. E. Kreischer, T. L. Grimm, W. C. Guss, A. W. Mobius, and R. J.
1
Temkin, Phys. Fluids B 2, 640 ~1990!.
Gyrotron Oscillators: Their Principles and Practice, edited by C. J. Edge- 18
T. L. Grimm, K. E. Kreischer, and R. J. Temkin, Phys. Fluids B 5, 4135
combe ~Taylor and Francis, London, 1993!. ~1993!.
2
K. Felch, M. Blank, P. Borchard, T. S. Chu, J. Feinstein, H. R. Jory, J. A. 19
K. E. Kreischer, M. Blank, B. G. Danly, T. Kimura, and R. J. Temkin
Lorbeck, C. M. Loring, Y. M. Mizuhara, J. M. Neilson, R. Schumacher,
IEDM Technical Digest ~Institute of Electrical and Electronics Engineers,
and R. J. Temkin, IEEE Trans. Plasma Sci. PS-24, 558 ~1996!.
3 New York, 1994!, IEEE Cat. No. CH35706, pp. 243–246.
M. Sato, T. Shimozuma, S. Kubo, Y. Takita, Y. Tsubokawa, K. Ohkubo, 20
S. Y. Cai, T. M. Antonsen, G. Saraph, and B. Levush, Int. J. Electron. 72,
T. Kuroda, H. Huey, and H. Jory, Fusion Eng. Design 26, 287 ~1995!.
4
G. G. Denisov, A. N. Kuftin, V. I. Malygin, N. P. Venediktov, D. V. 759 ~1992!.
21
Vinogradov, and V. E. Zapevalov, Int. J. Electron. 72, 1079 ~1992!. M. Blank, K. Kreischer, and R. J. Temkin, IEEE Trans. Plasma Sci. PS-
5
M. Pain, G. Tonon, T. M. Antonsen, Jr., M. Q. Tran, T. M. Tran, D. R. 24, 1058 ~1996!.
Whaley, E. Borie, C. Gruber, A. Mobius, B. Piosczyk, J. Pretterebner, M.
22
Y. Y. Lau, J. Appl. Phys. 61, 36 ~1987!.
Thumm, A. Wien, A. Dubrovin, P. Garin, E. Giguet, J. M. Krieg, and H.
23
J. M. Barid and W. Lawson, Int. J. Electron. 61, 953 ~1986!.
24
G. Mathews, Proceedings of the 18th Symposium on Fusion Technology, K. E. Kreischer, B. G. Danly, P. Woskoboinikow, W. J. Mulligan, and
Karlsruhe, Germany, 1994 ~Elsevier, Amsterdam, 1995!, pp. 489–492. R. J. Temkin, Int. J. Electron. 57, 851 ~1984!.
6 25
G. Dammertz, O. Braz, C. T. Iatrou, M. Kuntze, A. Mobius, B. Poisczyk, D. R. Whaley, M. Q. Tran, T. M. Tran, and T. M. Antonsen, Jr., IEEE
and M. Thumm, IEEE Trans. Plasma Sci. PS-24, 570 ~1996!. Trans. Plasma Sci. PS-22, 850 ~1994!.

1914 Phys. Plasmas, Vol. 4, No. 5, May 1997 Kreischer et al.

Downloaded 28 Jul 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions