3142 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 43, NO. 9, SEPTEMBER 2015

Strapped Magnetron Performance Affectedby Dielectric Material Filling

Sandeep Kumar Vyas, Shivendra Maurya, Rajendra Kumar Verma, and Vindhyavasini Prasad Singh

Abstract— This paper describes the output performance ofa strapped vane type 2.450 ± 0.030-GHz magnetron in thepresence of a dielectric material. A 3-D particle in cell softwarecomputer simulation technology has been used for this study.The side resonators (cavities) of the magnetron are partiallyfilled with a low-loss dielectric material (εr = 11.7 andtanδ = 3.5 × 10−6). The efficiency and output power of themagnetron have been increased by 2% and 10%, respectively,by dielectric material filling. It has been found that oscillationstart-up time of magnetron is not directly affected by dielectricmaterial filling as described in previous research papers.In addition, it has also been found that magnetron outputperformance remains almost the same for the lossless and thelow-loss dielectric material.

Index Terms— Dielectric filling, industrial magnetron, strapand vane resonator, virtual prototyping.

I. INTRODUCTION

AMAGNETRON is a high-frequency high-efficiencylow-cost microwave device. High-power continuous

wave (CW) magnetrons are being used in different typesof domestic, commercial, and industrial microwave ovensfor food and mineral processing. Many scientific projectslike space solar power system (SSPS) require high-efficiencylow-noise phase-locked magnetrons [1], [2]. In the SSPS,a very large number of magnetrons are used. Even a smallincrement in efficiency has a valuable importance. Somestudies of noise performance and phase stability have alreadybeen reported in [3]–[5]. This paper is mainly oriented towardmagnetron efficiency enhancement. Some efficiency enhance-ment techniques for relativistic magnetrons, like dielectricloading, axial power extraction, and priming techniques havebeen reported in [6]–[8].

The efficiency enhancement techniques for CW nonrela-tivistic magnetrons have not been reported so far to the bestof author’s knowledge. This paper describes the dielectricfilling effect on oscillation start-up time, output power andefficiency of a CW magnetron. A step by step simulationstudies have been carried out to study the output performance

Manuscript received December 20, 2014; revised May 12, 2015 andJune 26, 2015; accepted August 3, 2015. Date of publication August 18, 2015;date of current version September 9, 2015. This work was supported by theCouncil of Scientific and Industrial Research, New Delhi. (Correspondingauthor: Sandeep Kumar Vyas.)

The authors are with the Council of Scientific and Industrial Research-Central Electronics Engineering Research Institute, Pilani 333 031, India,and also with the Academy of Scientific and Innovative Research,New Delhi 110025, India (e-mail: sandeepvyas19@gmail.com; smaury@ceeri.ernet.in; rajendra.verma89@gmail.com; vvpsingh@ceeri.ernent.in).

Color versions of one or more of the figures in this paper are availableonline at http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TPS.2015.2465139

Fig. 1. Cut view of the basic magnetron resonator.

of the CW magnetron. Fig. 1 shows the 3-D model of themagnetron analyzed in this paper (now on defined as basicmodel). A cold test model of 10-vane 10 kW 2.45-GHz CWmagnetron has been developed and tested for verifying theaccuracy of simulation. It is found that the cold measurementresults very closely match the simulated results. The samemagnetron model has been used for dielectric filling effectstudy in various conditions like alternate resonators withpartial dielectric filling, all resonators with partial dielectricfilling, and dielectric loss effect etc. In [7] and [8] for relativis-tic magnetron, simulation studies have been performed withlossless (tanδ = 0) dielectric material. These papers show thatthe oscillation start-up time of magnetron reduces by dielectricfilling. In this paper, we found that the start-up time is unaf-fected by the dielectric loading if the operating points of basicmodel and dielectric filled magnetron are same. This paperhas been divided into four sections including the introduction.Section II briefly describes the basic magnetron simulationand its experimental evaluation. The effect of dielectric fillingon output performance of the magnetron through simulationshas been presented in Section III, followed by the conclusionin Section IV.

II. SIMULATION AND EXPERIMENTAL EVALUATION

OF BASIC MAGNETRON

This section presents the electromagnetic simulation andexperimental evaluation of basic model (unfilled resonators)of 10-kW CW strapped magnetron. The dimensional andoperating parameters of this magnetron are given in Table I [9].The simulation model, modeled in computer simulationtechnology (CST) microwave studio, and the electricfield vector plot for π-mode of oscillation are shown inFig. 2(a) and (b), respectively.

Fig. 2(b) shows that the π-mode frequency is 2.476 GHz.The magnetron resonator parts, experimental cold test setupwith assembled magnetron, and its experimental results are

0093-3813 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

VYAS et al.: STRAPPED MAGNETRON PERFORMANCE AFFECTED BY DIELECTRIC MATERIAL FILLING 3143

TABLE I

PARAMETERS OF 10-kW CW (BASIC MODEL) MAGNETRON

Fig. 2. (a) CST generated simulation model. (b) Electric field vector plotfor π-mode of oscillation.

shown in Fig. 3(a)–(c), respectively. In the cold test setup, themagnetron is connected to the vector network analyzer port bya coaxial transition. The value of voltage standing wave ratioof coaxial taper transition is less than 1.1 in the frequencyrange of interest [10]. The measured frequencies of π-modeand next higher mode are 2.445 and 3.735 GHz, respectively,as shown in Fig. 3(c). These are in quite good agreement withthe simulation results.

III. SIMULATION OF DIELECTRIC FILLING EFFECT

A. PIC Simulation of Basic Model

The particle in cell (PIC) simulation helps to simulate theoutput performance of magnetron by simulating the interactionbehavior of the RF circuit with the electron beam [10].The operating point for the π-mode of oscillation in themagnetron was obtained from the Hull cutoff voltage (Vc)and the Buneman–Hartree voltages (VT ) plots as a functionof the external magnetic field (B). Equations (1) and (2) areHull and Hartree equations, respectively [11]. The calculatedoperating point is marked at the anode voltage of 12.7 kV andthe applied axial magnetic field of 0.19 T as shown in Fig. 4.

The oscillation starts at 100 ns by the electrons flowingtoward the anode marking the onset of bunching and resultingin spokes formation as shown in Fig. 5(a). Fourier transfor-mation of the output signal that gives the oscillation spectrumis shown in Fig. 5(b). From the oscillation spectrum, we get

Fig. 3. (a) Magnetron resonator parts. (b) Cold test setup with experimental10-kW magnetron anode. (c) Experimental result of resonant pattern.

Fig. 4. Magnetron operation domain.

the hot frequency for π-mode of oscillation is 2.466 GHz.Fig. 5(c)–(f) shows the time evolution of cathode emittedcurrent, cathode collected current, anode current, and outputpower, respectively. The anode current and output power arefound to be 1.21 A and 10.98 kW, respectively, with anefficiency of about 71.45%

Vc = eB2r2a

8m

[1 −

(rc

ra

)2]2

(1)

VT = πcr2a

nλ

(1 − r2

c

r2a

)B −

(2π2c2r2

a m

eλ2n2

)(2)

where e, m, c, n, and λ are charge of electron, mass ofelectron, velocity of light, mode number, and wavelength ofmode, respectively.

3144 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 43, NO. 9, SEPTEMBER 2015

Fig. 5. (a) Electron bunch (spokes) formation at 300 ns. (b) Fourier transformof the RF output signal. (c) Time evolution of cathode emitted current.(d) Time evolution of cathode collected current. (e) Time evolution of anodecurrent. (f) Time evolution of RF output power at the absorption port.

Fig. 6. (a) Eigen mode simulation model. (b) Electric field vector plot ofπ -mode with partial dielectric filling.

B. N/2 Vanes Dielectric Filled Magnetron Model

In this simulation model, alternate resonators (N /2) of thebasic model are partially filled with Au-doped silicon asshown in Fig. 6(a). The dielectric constant and loss tangent ofAu-doped silicon are 11.7 and 3.5e-6, respectively. The fillingfactor is taken to be about 30% of vane length. The inner radiusof dielectric annular sector (rd) is 16 mm. Before simulation ofnonlinear beam-wave interaction, electromagnetic simulationusing the proposed geometry has been carried out. The eigenmode simulation model and electric field vector plot for thiscase are shown in Fig. 6(a) and (b), respectively.

From Fig. 6(b), it is clear that the π-mode frequency for thiscase is 2.423 GHz, which is 53 MHz down from that of the

Fig. 7. (a) Electron bunch formation at 400 ns. (b) Time evaluation of outputRF power. (c) Time evolution of anode current. (d) Time evolution of cathodeemitted current.

basic model. After electromagnetic simulation, PIC simulationhas been carried out with an anode voltage of 12.7 kV and amagnetic field of 1900 G. Different output parameters obtainedfrom PIC simulation are shown in Fig. 7(a)–(d).

It can be concluded from Fig. 7(b) that oscillation start-uptime in this case has been reduced by 50 ns compared withbasic model. The output power and anode current are alsoincreased as shown in Fig. 7(b) and (c), respectively. Thedc impedance of the magnetron has been decreased. ThePIC simulation results for this case have been summarizedin Table II (S. No. 2). It is well known that the magnetronoperating point (anode voltage and magnetic field) is decidedby Hull cutoff parabola and Hartree diagram and Hartreeline is a function of frequency. When the magnetron sideresonators are filled with the dielectric material, its frequencydecreases, and consequently, the Hartree line has been shiftedtoward the x-axis as shown in Fig. 8. Now, we have chosen anew operating point (P2) with the same percentage separationwith shifted Hartree line as the previous operating point (P1)had with the original Hartree line. Again the PIC simulationhas been carried out with the new operating point (P2) withan anode voltage of 12.7 kV and a magnetic field of 1932 G.

The summary of PIC simulation results with new operatingpoint has been given in Table II (S. No. 3). FromTable II (S. No. 3), it is clear that the oscillation start-up timewith this operating point (P2) is almost the same as that ofthe basic model. Hence, it can be predicted that the oscillationstart-up time does not depend on dielectric material filling.To confirm this, we changed the cold frequency of basic modelto 2.425 GHz by increasing strap thickness by 0.16 mm. Nowthis magnetron has been simulated for both the operatingpoints (P1: 12.7 kV, 1900 G and P2: 12.7 kV, 1932 G).The results of these PIC simulations are givenin Table II (S. Nos. 4 and 5). The comparison of allresults so far from Table II, confirmed that oscillation time

VYAS et al.: STRAPPED MAGNETRON PERFORMANCE AFFECTED BY DIELECTRIC MATERIAL FILLING 3145

TABLE II

COMPARISON OF OSCILLATION TIME FOR UNFILLED AND DIELECTRIC FILLED MAGNETRONS VA

Fig. 8. Shift in Hartree line and correspondingly chosen new operating pointdue to dielectric material filling.

does not depend on the dielectric filling; instead, it dependsonly on the operating point of the magnetron.

C. N Vane Dielectric Filled Magnetron Model

In this case, we have filled all the 10 side resonators ofbasic model with the same dielectric material and same fillingfactor as described in Section III-B as shown in Fig. 9(a). Theelectric field vector plot in this case for π-mode of oscillationis shown in Fig. 9(b).

It is found that in the case of N vane dielectric filledmodel the cold π-mode frequency has been shifted downwardto 2.375 GHz, which is 99 MHz lower than that of thebasic model. Now we have calculated the required magneticfield for shifted frequency, i.e., 2.375 GHz as was calculatedin Section III-B and it was found to be 1962 G.

Fig. 9. (a) Eigen mode simulation model. (b) Electric field vector plot ofπ -mode.

Next, the PIC simulation has been carried out for thiscase with two operating points (12.7 kV and 1900 G) and(12.7 kV and 1962 G). The summary of the PIC simulationresults for both the operating points has been givenin Table II (S. Nos. 6 and 7). From Table II (S. Nos. 6 and 7),once again it can be concluded that oscillation start-up time isaffected by operating point. Hence from the simulation studiescarried out so far it may be concluded that the partial dielectricfilling in side resonator of a 10-vane 2.45-GHz CW magnetronchanges/reduces the π-mode resonance and has no effect onthe oscillation start-up time.

D. Dielectric Filled Magnetron Keeping FrequencyFixed at 2.475 GHz

In this paper, we have filled the dielectric material in alter-nate cavities/all cavities and π-mode oscillation frequencieshave been maintained at, or close to, that of the basic model,i.e., 2.475 GHz, by slightly changing the strap dimensions andeffect of dielectric filling on efficiency (η) has been observed.

In addition, a 10-vane dielectric filled case consideringlossless dielectric has also been studied. The PIC simulationresults are tabulated in Table III. Dimensional parameters are

3146 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 43, NO. 9, SEPTEMBER 2015

TABLE III

PIC SIMULATION OF MODIFIED DIELECTRIC FILLED

MAGNETRON AT 12.7 kV AND 1900 G

the same as given in Table I except radial gap between strapshas been increased by 0.2 mm for 5-vane filled case and0.6 mm for 10-vane filled case. In the 10-vane filled case,the strap thickness has also been increased by 0.02 mm toachieve the basic π-mode frequency, i.e., 2.475 GHz.

From Table III, it can be concluded that in the caseof the 10-vane dielectric filled case the efficiency of the10-KW 2.45-GHz CW magnetron has been increasedby ∼2% compared with that of the basic model. It was alsofound that low-loss dielectric filling is as effective as losslessdielectric filling. A host of low-loss dielectric available willbe used in practical magnetrons.

IV. CONCLUSION

CST particle studio has been used to examine the outputperformance of a 10-KW 2.45-GHz CW magnetron in thepresence of dielectric material in the side resonators. It wasfound that in the presence of dielectric material the efficiencyof magnetron has been increased by ∼2%. It is also foundthat for this type of particular magnetron oscillation start-uptime is not affected by dielectric filling. The output power hasbeen increased by 10% in the case of lossless dielectric filledcavities compared with that of basic model.

Further study is required to fix dielectric filling factor andsuitable dielectric material.

ACKNOWLEDGMENT

The authors would like to thank the director, CSIR-CEERI,Pilani, for granting permission to publish this paper.

REFERENCES

[1] T. Mitani, H. Kawasaki, N. Shinohara, and H. Matsumoto, “A study ofoven magnetrons toward a transmitter for space applications,” in Proc.IVEC, Rome, Italy, Apr. 2009, pp. 323–324.

[2] W. C. Brown, “Satellite power system (SPS) magnetron tube assessmentstudy,” NASA, Wasington, DC, USA, Tech. Rep. 3383, Feb. 1981.

[3] A. C. Dexter et al., “First demonstration and performance of an injectionlocked continuous wave magnetron to phase control a superconductingcavity,” Phys. Rev. ST Accel. Beams, vol. 14, no. 3, p. 032001, Mar. 2011.

[4] V. B. Neculaes, R. M. Gilgenbach, Y. Y. Lau, M. C. Jones, andW. M. White, “Low-noise microwave oven magnetrons with faststart-oscillation by azimuthally varying axial magnetic fields,” IEEETrans. Plasma Sci., vol. 32, no. 3, pp. 1152–1159, Jun. 2004.

[5] T. Mitani, N. Shinohara, H. Matsumoto, M. Aiga, N. Kuwahara, andT. Ishii, “Noise-reduction effects of oven magnetron with cathode shieldon high-voltage input side,” IEEE Trans. Electron Devices, vol. 53, no. 8,pp. 1929–1936, Aug. 2006.

[6] M. I. Fuks and E. Schamiloglu, “70% efficient relativistic magnetronwith axial extraction of radiation through a horn antenna,” IEEE Trans.Plasma Sci., vol. 38, no. 6, pp. 1302–1312, Jun. 2010.

[7] S. Maurya, V. V. P. Singh, and P. K. Jain, “Study of output performanceof partially dielectric loaded A6 relativistic magnetron,” IEEE Trans.Plasma Sci., vol. 40, no. 4, pp. 1070–1074, Apr. 2012.

[8] S. M. A. Hashemi, “Dielectric cavity relativistic magnetron,” Appl. Phys.Lett., vol. 96, no. 8, pp. 081503-1–081503-3, Feb. 2010.

[9] S. Vyas, S. Maurya, and V. V. P. Singh, “Synthesis and analysis ofstrap & vane resonator of an efficient 10 kW CW magnetron—A designapproach,” CEERI, Pilani, India, Internal Rep. MWT/RR-1/2014, 2014.

[10] S. K. Vyas, S. Maurya, and V. V. P. Singh, “Electromagneticand particle-in-cell simulation studies of a high power strap andvane CW magnetron,” IEEE Trans. Plasma Sci., vol. 42, no. 10,pp. 3373–3379, Oct. 2014.

[11] R. S. H. Boulding, The Resonant Cavity Magnetron. New York, NY,USA: Van Nostrand, 1952, Ch. 7.

Authors’ photographs and biographies not available at the time of publication.

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具