Search

Investigations of magnetic field diffusion and plasma resistivity in 12-cm-diameter triple-gas-puff Ar Z-pinch implosions were carried out by using planar laser-induced fluorescence (PLIF), a laser shearing interferometer (LSI), and a laser wavefront analyzer (LWA) on a 3.5-MA 200-ns generator. The PLIF measurements gave the initial Ar gas distributions. The implosion velocity and electron density profiles were measured from LWA and/or LSI. From these, the implosion plasma sheath thickness, ion density, mean ion charge states, temperatures, and implosion velocity are obtained, which allows us to calculate the classical plasma resistivity. A 1-D analytic magnetic field diffusion model is constructed and used to predict the imploding plasma sheath thickness and its resistivity. Based on comparisons of the experimental measurements and the diffusion model prediction, we found out that plasma resistivity is enhanced by the cross-field diffusion above the classical value, as high as 60 times the Spitzer's value. Details are given in this paper.

We have applied the technique of acetone laser-induced fluorescence (LIF) to the measurement of gas distributions from axisymmetric supersonic nozzles used to produce loads for z-pinch plasma radiation sources. Typical peak particle densities are ∼1017 particles/cm3 for loads imploded on the Double-EAGLE facility. The experimental approach uses a pulsed laser (266 nm wavelength, 2.2 mJ per pulse, 5 ns pulse width, and 3×107 W/cm2 intensity) to obtain a snapshot along a chord through the center of the gas density distribution at an arbitrary axial distance, z, from the nozzle exit. We report measurements at 4.3 and 20.0 mm from the exit of the nozzle for comparison with previous measurements. We find acceptable agreement between LIF and laser interferometer measurements. Strengths of the LIF approach include simplicity of implementation and high radial spatial resolution.

Hardware and software have been developed for recording and displaying accurate image and spectral data produced by z-pinch plasma radiation sources at the Double-EAGLE facility at Maxwell Physics International. Desktop computers are used to acquire the data, analyze it, and display and print the results. Of the four charge-coupled device (CCD) image recording systems implemented, two record x rays directly and two record optical light emission from electron–excited phosphors. The CCD systems required careful shielding to allow them to operate in the harsh radio frequency noise environment. During a series of shots at the SATURN facility at Sandia National Laboratories, the quality of a keV x-ray spectrum recorded directly with a CCD compared well with an equivalent spectrum recorded with 2497 film.

Solid-fill nozzles for long-implosion Z-pinch experiments to produce argon K-shell x rays (photon energy >3.1 keV) have been developed. With a 7 cm diam nozzle, which is appropriate for a 200 ns driver, stable implosions at 180 ns and 4 MA have produced peak argon K-shell yields exceeding 15 kJ. As previously seen with short (∼100 ns) implosion times, the K-shell yield scales as the fourth power of peak current, I4. Limited testing with a 10 cm nozzle, which is appropriate for a >250 ns driver, has also achieved a stable implosion.

Timely estimates of z-pinch plasma conditions can aid in the optimization of plasma radiation source K-shell emission. For example, a shell-on-shell argon gas puff has many parameters, such as the axial gas density profile, inner to outer shell mass ratio, total mass, and pinch load length, which can be varied on a shot by shot basis. At the Double-EAGLE facility at Maxwell Physics International, digital cameras have been implemented to routinely record pinhole images, x-ray spectra, and axially resolved streak images. A computer algorithm has been implemented to take as input the data (pinch diameter, Ar Lyα to Heα line ratio, and K-shell power) and output the plasma parameters (density, electron temperature, and K-shell-emitting mass) as a function of z-pinch axial location. We describe the algorithm and present results for three different z-pinch loads.

High-current (1–3 MA) reflex triodes are being developed as intense pulsed sources of moderate energy (200–1200 kV endpoint) bremsstrahlung. The triode consists of three coaxial electrodes: two hollow cylindrical cathodes and a flat anode foil between them. The grounded cathodes emit electrons that produce bremsstrahlung in the positive high voltage anode. The anode is made from thin tantalum (small fraction of the electron range) to minimize attenuation of the low energy portion of the x-ray spectrum. The electrons must make many passes through the tantalum to efficiently lose their energy and produce x-rays. This process is called reflexing. The triode configuration is thought to be the best way to accomplish this.

Summary form only given. Characterization of implosion kinematics is of interest in understanding and optimizing plasma radiation source (PRS) K-shell yield. It also serves to provide input parameters to the modeling of PRS pinch conditions. A measure of the sensitivity of the PRS implosion kinematics to changes in initial conditions is the time and axial dependence of the Z-pinch assembly on-axis and its associated K-shell yield. A relatively simple and routine PRS diagnostic is the so-called zipper array, which can characterize both the time and axial dependence of the pinch assembly (zippering) and the pinch x-ray yield. The zipper array is essentially a one-dimensional (1D), time-resolved imager, typically consisting of a linear array of X-ray power sensors (photo-conductive detectors: silicon diodes or diamonds), viewing the pinch through a transverse slit, with appropriate pre-detector filtering selecting the X-ray detection range of interest. We report here a study on the time dependent axial characterization of the pinch assembly and the K-shell yield for argon PRS shots taken with two different current drives, differing both in magnitude and waveform shape, and with implosion times in excess of 200 ns. The first current driver reached a peak current plateau of about 3.5 MA in 100 ns, which persisted at nearly constant current for an additional ~100 ns. The second driver delivered a nominally linear current ramp exceeding 6 MA in 300 ns. The study characterizes the final pinch assembly under both fixed and varied initial conditions. For nearly identical initial conditions, the data generally show a significant degree of reproducibility in the relative zipper timing, but with some variation in the K-shell yield at a given axial position. Changes in initial conditions can have a significant effect. For example, shots without pre-ionization changed the axial time dependence of final pinch assembly and axial distribution of K-shell yield.

A fast Marx (220 kJ, 450 nH, 675 kV) has been developed as a driver for an electron beam generator. This generator uses an inductive store with an opening switch to compress and amplify the power pulse. The 220 kJ Marx consists of four parallel 55 kJ Marxes to minimize the inductance. The 220 kJ Marx has been used to develop vacuum opening switches. These opening switches have demonstrated up to 22% efficiency and pulse compression of 5 to 1. Two switch concepts have been successfully tested as electron beam diodes on this Marx bank. The energy delivered to diode, the output voltage and power produced by this new high power generator will be presented. This generator serves as the proof-of-principle demonstration of inductive store/opening switch technology as an alternative to conventional capacitive store waterline technology.

The authors are developing an L-band (1.3 GHz) high current relativistic klystron amplifier for both long pulse ({ge} 500 ns) and high repetition rate ({ge} 200 pps) capabilities. In repetitive operation, it has produced 3.3 kW of average power in 80 ns FWHM pulses of 250 MW peak power at 200 pps. They are currently transferring this same klystron to a single-shot, long-pulse driver to demonstrate its operation with a {ge} 500 ns electrical pulse.

We developed four types of GaAs PCDs: bremsstrahlung (two types), soft X-ray, and gamma for pulsed reactors. GaAs PCDs are advantageous in applications that require fast response, high dose rate, and/or neutron insensitivity. Desired detector sensitivity and response were obtained through modification of carrier lifetime by neutron irradiation, size of the sensing element, and orientation of the sensing element. The GaAs PCDs were used to characterize pulsed radiation sources in over 10 different tests and the measured data from these tests are presented. GaAs PCDs (E7 - E11 rad(Si)/s) for use with pulsed bremsstrahlung sources were tested at DTRA Radiation sources, Linac (Idaho Accelerator Center), Linac (White Sands), hybrid radiation source (NRL), GAMBLE II (NRL), and HERMES III (Sandia). GaAs PCDs for gamma characterization in pulsed reactors (E6 - E7 rad(Si)/s) were calibrated with a Linac (White Sands), checked for high dose rate (E12 rad(Si)/s) response at HERMES III, checked for low dose rate response (5-500 rad(Si)/s) and stability at a Co-60 source (Sandia), and checked for stability, sensitivity, and residual radioactivity at the ACCR reactor (Sandia). Soft X-ray (1 to 15 keV) GaAs PCDs (E2-E6 Watts/cm2) were used to measure the argon plasma free-bound continuum temperature from a plasma radiation source (PRS) and compared with the plasma temperature obtained from line ratio measurements.

Summary form only given. Because the energy coupling and the X-ray output of Z-pinch plasma radiation sources are determined by the implosion history, measurements from the initial gas phase, through the MHD implosion phase, to the final pinch phase are needed. Many diagnostics have been developed for the final pinch phase. We have developed a planar laser induced fluorescence (PLIF) technique to characterize the gas puff in the initial phase and demonstrated a laser shearing interferometer (LSI) approach for the implosion and pinch phases on a high current accelerator. These laser based instruments have the advantages of being user-friendly and giving the data quickly. Using the PLIF method, the 2-D gas density and flow velocity profiles of the loads were obtained with sub-mm spatial resolutions for the conditions used in the Z-pinch experiments. From the LSI, the shape of the imploding plasma was determined and the wavelengths of the Rayleigh-Taylor instability were measured. We have also found that the pinch plasma sizes are about twice that measured from the K-shell X-ray pinhole images. The combination of the PLIF and (LSI) measurements enable us to better understand the implosion dynamics and to provide important information needed for simulation codes

Summary form only given. A number of diagnostics have been developed to measure the argon K-shell (~3 keV) X-ray emitting volume and spectrum. However, only ~20% of the load mass participates in the K-shell emission and less than 20% of the total emission falls in that energy band. We have made measurements to determine both the radial emission profile and the emitted spectrum in the XUV (0.1

Summary form only given. The technique of Planar Laser Induced Fluorescence (PLIF) has been applied to characterize the initial gas distributions produced by 12 cm diameter nozzles, as used on recent Decade Quad and Double-EAGLE Ar Z-pinch experiments. This information is required to understand and simulate the implosion dynamics. A 266 nm, /spl sim/5 ns, /spl sim/5 cm line-focused laser beam passes through the Ar gas puff, which is mixed with a few percent acetone by volume. A gated intensified CCD camera views the fluorescence of the excited acetone molecules from the side. The acetone emission has a short, intense pulse, coincident with the exciting laser pulse, and a weak, slowly decaying phosphorescence. The gas density profiles are obtained from the fluorescence images recorded coincident with the laser. Flow velocities of the gas puff are obtained by capturing the images of the phosphorescence at several delay times relative to the laser pulse. The distribution of the clusters is determined by comparing the PLIF signal with 2-D Rayleigh scattering from the gas puff. These measurements are compared with computer simulations to benchmark and refine the 2-D fluid dynamics models.

The Decade half X-ray radiation simulator will combine the outputs of eight synchronized modules to produce a current of up to 13 MA delivered to a short circuit load in /spl sim/300 ns. Decade is located within the Decade Radiation Test Facility (DRTF) at the Arnold Engineering Development Center, Tullahoma, TN. Decade's output will be used for nuclear weapons effects testing; the DRTF will reach initial operational capability (IOC) in 2007. DoD's X-ray radiation simulator R&D program is focused on maximizing 3.1 keV X-ray fluence from Decade's argon z-pinch. Electric current risetime will be approximately a factor of 3 longer than typically used to drive plasma radiation source (PRS) loads. Scaling to longer current risetime is advantageous in terms of reduced driver cost and complexity. The use of longer pulse drivers, however, requires an increase in the initial diameter of the z-pinch gas puff. The unique challenge is to extend the success with 100 ns current risetime simulators into the realm of the 300 ns current risetime of Decade. As the diameter is increased, increased asymmetry and instability can limit the ability of the load to produce K-shell radiation efficiently. These effects were largely un-quantified until technical investigations were conducted under the auspices of the simulator R&D program. This paper discusses progress, activities and issues in developing the large diameter z-pinch for Decade.

Summary form only given, as follows. In a previous research, an azimuthally symmetric UV source was used to preionize the argon gas puffs in the DM2 plasma radiation source (PRS) experiment. It indicated that the preionization helped to improve the PRS K-shell yield and reduce shot to shot variations. It also revealed that the issues of survivability and replaceability must be addressed for such a preionization source to be used in super-power generator environments. Based on the evaluation of the previous preionizer, a robust new preionizer has been fabricated and fielded on Double-EAGLE during the PRS experiment. Design of the new preionizer and its performance on Double-EAGLE will be presented. Comparison with the previous preionizer will also be discussed.

We have created large diameter, long implosion time, PRS loads that produce a stable pinch and a narrow X-ray pulse. Operation scaled successfully with current from 2 MA on DM2 to 6 MA on Saturn. The K-shell yields are comparable to those produced by short-pulse operation at a similar current. This approach, therefore, allows substantial savings to be realized in the pulsed-power driver cost, complexity and risk.

DECADE Module 2 (DM2) implodes argon gas-puffs in 200-300 ns from initial radii of 2.5-5 cm to generate K-line radiation. An important issue for these experiments is the lower-than-expected K-line radiation associated with such large radius implosions. Preionization may improve the implosion quality by improving the symmetry of the initial current flow. A preionizer is described based on photoionization by ultraviolet (UV) illumination from a flashover source. The preionization is diagnosed using a high-sensitivity interferometer. This UV source produces 1-10% ionization of the outer periphery of the argon gas distribution prior to the arrival of the flashboard plasma. DM2 experiments indicate increased, more reproducible K-shell yields when this preionization system is incorporated.

Summary form only given, as follows. We are continuing the development of a low density plasma opening switch (POS) on the Double-EAGLE pulsed power generator, running in a 250 ns mode. The goal is to shorten the pulse and efficiently drive a=100 ns imploding PRS source (an aluminum wire array). This will provide a proof-of-principle demonstration of an alternative approach to long implosion PRS sources that are being developed for Decade Quad and the next generation of pulsed power drivers. In our previous experiment, the upper and lower current paths of Double-EAGLE strip-plate geometry were connected upstream of the POS to test the operation of a single-sided low density switch at linear current densities required for Decade Quad. Our present efforts involve using the lower inductance geometry of a double-sided POS and a post-hole convolute connecting the current paths downstream of the POS, driving either an inductive shun or an aluminum wire array. These tests will attempt to demonstrate high opened POS impedance and minimum electron sheath flow ("launched electron flow") downstream of the POS. We will discuss the interactions of the two sides of the POS and of the POS with the loads. The implications for Decade Quad will also be addressed.

Summary form only given, as follows. We have applied the technique of acetone laser-induced fluorescence (LIF) to the measurement of gas distributions from axisymmetric, supersonic nozzles used to produce loads for z-pinch plasma radiation sources. Typical peak particle densities are /spl sim/1% of an atmosphere at standard temperature. A pulsed laser is used to obtain a snapshot along a chord through the center of the gas density distribution at 4.3 and 20.0 mm from the exit of the nozzle. We find good agreement between LIF and laser interferometer measurements. Strengths of the LIF approach include simplicity of implementation and high radial spatial resolution.

Summary form only given, as follows. The Doppler broadening of ion line profiles emitted by a Z-pinch plasma provides information about the thermalization of the implosion kinetic energy and the radiation efficiency of the pinch. Measurement of these line profiles is often complicated by source broadening in the instrument and opacity broadening of the emitted radiation. A high resolution concave crystal spectrometer in the Johann geometry was used to record the time averaged spectra of optical thin trace elements in the load. An imaging slit provided radially resolved but axially averaged spectra. Source broadening is overcome by imaging monochromatic X-rays emitted from all axial pinch positions to one point on the Rowland circle. The recording film is located on the Rowland circle whose radius is half the crystal radius and whose circumference is tangent to the center of the crystal. The spectral resolution of the spectrometer, /spl lambda///spl Delta//spl lambda//spl sim/1000, was measured with K radiation generated by a 10 keV electron beam incident on a Si or Al target. Measurements are reported for Al wire loads on both the Saturn and Double Eagle accelerators in both a short current pulse mode (60-100 ns) and a long pulse mode (125-225 ns). On Double Eagle we observed ion temperatures in the 3-5 keV range in the short pulse mode and 20-30 keV in the long pulse mode. On Saturn, on the other hand, we observed the larger values for both current modes.

Summary form only given. The next generation of pulsed power drivers will likely have much longer current pulse lengths than the present generation, 200-300 ns as compared to 50-100 ns. This presents the problem of efficiently connecting FRS loads, which have been optimized for the shorter pulses, to the longer pulse drivers. One approach has been to develop PRS sources that can be directly driven by the longer current pulse without suffering from the Rayleigh-Taylor instability. While this has met with some success, the radiation efficiency (kJ/MA) is only /spl ap/50% of that produced in short pulse. Another approach that we investigated, and report upon here, is to use a large volume, low density, plasma opening switch to conduct the full current for /spl ap/100 ns and then open, transferring the current to the PRS load. The resultant drive pulse seen by the PRS is thus comparable to that produced by short pulse drivers. For the experiments reported here, the Double-EAGLE current pulse was lengthen to 250 ns by shorting out switches in the waterlines.

The K-shell yield performance of Titan double-shell argon gas puff loads imploded on the Z machine and Decade Quad (DQ) is compared to the performance of Double-Eagle loads employing the same nozzle configuration. Specifically, the K-shell yields obtained on the Z and DQ machines are compared with I4, I2, and I0 current scaling projections made from Double-Eagle yields, where I is the peak load current. This analysis allows initialization and multidimensional effects to be factored out of the scaling behavior. This projection analysis is useful for evaluating argon gas-puff load performance because it relates I4, I2, and I0 K-shell yield scaling transitions that a load, of a given specific energy, is predicted to undergo as the total load mass increases. Our analysis shows that the maximum K-shell yields were near optimum for the nozzle configuration used and the energy coupled to the load in the Z and Decade Quad experiments. At present it is not understood why an additional 22 nH of inductance for DQ, 6 nH for the Z machine, and 8 nH for Double-Eagle is required in their respective equivalent circuit models to reasonably match measured peak currents and implosion times. Note, we do not think that there is extra inductance in the real circuit; it is included in the modeled equivalent circuit as a method to account for the possible anomalous effects that are reducing the current from its expected value. These effects may include: multidimensional effects, current losses, and anomalous resistivity.

Summary form only given, as follows. We have continued our development and testing of uniform-fill nozzles for long-implosion time PRS sources on the 4 MA Double-EAGLE generator at Maxwell Physics International. By adjustment of switch gaps in the waterline, the nominal 100 ns electrical pulse of Double-EAGLE can be extended to 200 or 250 ns with little decrease in peak current. We have used these modes to test 7 and 10 cm diameter uniform fill gas puffs in well diagnosed experiments. The 7 cm nozzle generated a K-shell yield in excess of 15 kJ in a 16 ns FWHM pulse at 188 ns implosion time and demonstrated I/sup 4/ scaling to the testing limit of 4 MA. The final implosion diameter was about 2 mm. The compression ratio was therefore better than that achieved in short pulse operation (2.5 cm initial shell diameter to 1 mm final pinch diameter). We will discuss these Z-pinch results, the 10 cm nozzle results, and features of the gas valve and nozzles employed to produce them.

The magnetically controlled plasma opening switch (MCPOS) is an advanced plasma opening switch that utilizes magnetic fields to improve operation. Magnetic fields always dominate terawatt, pulsed power plasma opening switches. For that reason, the MCPOS uses controlled applied magnetic fields with magnitude comparable to the self-magnetic field of the storage inductor. One applied field holds the plasma in place while energy accumulates in the storage inductor, then another applied field pushes the plasma away from the cathode to allow energy to flow downstream. Over a ten month period, an MCPOS was designed, built, and tested on DECADE Module 2 at Physics International. The peak drive current was 1.8 MA in 250 ns. The output parameters were up to 1 MA into an electron beam load. The radiation temporal pulse width averaged 60 nanoseconds full-width at half-maximum. The peak load voltage ranged from one to two megavolts. The experiments demonstrated efficient power flow through a long, low-impedance magnetically insulated transmission line between the magnetically controlled plasma opening switch and the load.

Recently, a new approach for efficiently generating K-shell x-rays in large-diameter, long-implosion time, structured argon gas Z-pinches has been demonstrated based on a “pusher-stabilizer-radiator” model. In this paper, direct observations of the Rayleigh–Taylor instability mitigation of a 12-cm diameter, 200-ns implosion time argon Z-pinch using a laser shearing interferometer (LSI) and a laser wavefront analyzer (LWA) are presented. Using a zero-dimensional snowplow model, the imploding plasma trajectories are calculated with the driver current waveforms and the initial mass distributions measured using the planar laser induced fluorescence method. From the LSI and LWA images, the plasma density and trajectory during the implosion are measured. The measured trajectory agrees with the snowplow calculations. The suppression of hydromagnetic instabilities in the “pusher-stabilizer-radiator” structured loads, leading to a high-compression ratio, high-yield Z-pinch, is discussed. For comparison, the LSI an...

A double-layer consists of two adjacent, oppositely-charged space-charge layers, and may occur near a material boundary, or within the plasma volume (See [1]-[3] for reviews of double-layer phenomena in laboratory and space plasmas). Here, we shall analyze a steady-state double-layer separating two plasmas of different densities and temperatures, and relate its length and potential drop to the plasma parameters. Several theoretical models have treated the double-layer as a region of high electric field and monotonic potential variation. The total potential drop is generally taken to be larger than the plasma thermal energy, and the electric field is confined to the double-layer, implying that the charge variation integrates to zero. Cold plasma, fluid, and kinetic theory approaches have been discussed [2].

Abstract A homogeneous and uniform cylindrical reservoir containing oil and gas is fractured vertically on completion and is produced at a constant bottom-hole pressure. The fracture has an infinite flow capacity, is of limited lateral extent and is bounded above and below by the impermeable strata defining the vertical extent of the reservoir. Results show that such a fractured reservoir can be represented by a reservoir of circular symmetry having very nearly the same production history. The well radius of this circular reservoir is about 1/4 the fracture length and is essentially the same as that obtained previously for a single fluid of constant compressibility.At the same value of cumulative oil production, gas-oil ratios of fractured reservoirs producing at constant terminal pressure are larger than those of reservoirs having no fractures. This leads to more inefficient use of the reservoir energy in fractured wells and results in lower reservoir pressures for the same cumulative oil production. The reduction in operating life due to fracturing a reservoir is not as great as that for a slightly compressible fluid. This difference can be accounted for by the lower reservoir pressure in the fractured reservoir and its adverse effect on the average mobility and compressibility of the oil. As anticipated. the reduction in operating life increases as the reservoir permeability decreases.The type of results presented in this report can be used to determine the economic attractiveness of fracture treatments per se, to select the initial spacing to be used in developing a field, and to compare the relative merits of fracturing available wells and in fill drilling. Introduction The effect of vertical fractures on a reservoir producing either an incompressible or a compressible liquid has already been discussed in the literature. Those results indicate that the production history of such a reservoir is essentially the same as that of a circular reservoir having an effective well radius of approximately one-fourth the fracture length.The present work reports on the effect of a vertical fracture on a reservoir producing two compressible fluids -oil and gas-by solution gas drive. Because of the empirical nature of the PVT and relative permeability data used to obtain the performance of such reservoirs, results can only be obtained numerically and with the aid of high-speed Computers.Since reservoirs lose their radial symmetry when fractured vertically, pressure and saturation can no longer be given only in terms of distance from the well. Two coordinates (such as x and y) must now be used to describe the pressure and saturation within the reservoir, and, since we are dealing with compressible fluids, time is also a variable. Thus the solution of a vertically fractured reservoir requires finding two unknowns (pressure and saturation) in two space variables (say x and y) and in time (t). Since no means are readily and generally available for solving such problems at the present time, we have used the results of previous work to approximate the effect of a vertical fracture on a reservoir producing both oil and gas by depletion.The purpose of the present work, then, is to investigate the possibility of using available numerical techniques (limited at the moment to one space variable) to study the two-space-variable flow behavior resulting from a vertical fracture. Results obtained in the course of this investigation are also reported and discussed.Input and output data of the numerical methods used are given in practical units: BOPD, feet, psi, cp, and md. Results are discussed first in terms of specific reservoir and crude properties and geometries. Later, dimensionless parameters are introduced in order to extend results to different values of some of the reservoir and fracture properties. IDEALIZATION AND DESCRIPTION OF THE FRACTURED SYSTEM It is assumed that a horizontal oil-producing layer of constant thickness and of uniform porosity and permeability is bounded above and below by impermeable strata. The reservoir has an impermeable, circular, cylindrical outer boundary of radius r . The fracture system is represented by a single, plane, vertical fracture of limited radial extent, bounded by the impermeable matrix above and below the producing layer (reservoir). It is assumed that there is no pressure drop in the fracture due to fluid flow. Fig. 1 indicates the general three-dimensional geometry of the fractured reservoir. Gravity effects and the effects of differential depletion resulting from variations in hydrostatic head (pressure) will be neglected. JPT P. 1119^

Gyrotron oscillators are of great interest as sources of high power mm wave radiation for electron cyclotron resonance heating and current drive in magnetic fusion research devices. However, as gyrotron power approaches the megawatt level, a very large, overmoded cavity must be employed in order to keep ohmic power losses in the cavity at an acceptable level, and the problem of mode overpopulation becomes severe. In this paper a novel approach for solving the problem by using a multiple, or complex cavity configuration is described. Prebunching is achieved in the first cavity while energy extraction takes place in the second. Experimental and theoretical studies for efficiency enhancement and mode competition reduction are presented for a 35 GHz, 350 kW gyrotron oscillator operating with a TE 01 /TE 04 cavity.