Fast and effective tuned coupling for mono-mode microwave power applicators Wojciech Gwarek Institute of Radioelectronics and Multimedia Technology Warsaw Univ. of Technology Warsaw, Poland Malgorzata Celuch Institute of Radioelectronics and Multimedia Technology Warsaw Univ. of Technology Warsaw, Poland Abstract A new concept for variation of coupling between a mono-mode applicator and a waveguide input is described. It provides a possibility for continuous tuning of the coupling factor, so that very good matching of the applicator is maintained while temperature modifies the loss factor of the heated material. The tuning is simple and much more convenient than a typical matching with a three-stub tuner. It is a very low-loss solution applicable also at very high power, even of the order of hundreds of kw. Index Terms Microwave power, Mono-mode applicator, Waveguide technology I. INTRODUCTION Mono-mode microwave power applicators are commonly used in research and in industry [1]. Usually they are composed of a cylindrical or cuboidal resonating region partially filled with the heated material and connected to the input rectangular waveguide through an iris. They have been subject to detailed studies like for example [2] and are offered as commercial products (eg. [3]). Typical resonant modes applied are: TM010 in a cylindrical cavity and TM110 or TM120 in a rectangular cavity. One of the main problems in such devices is proper matching of the applicator in the course the heating process. Fig. 1. Equivalent scheme of a resonator coupled to a waveguide. Consider the matching problem with a simple schematic diagram of Fig. 1. On the left hand side we have equivalent elements describing the resonator (L,C,G). The resonator separated from the outside world has its proper (unloaded) Q u =ωc/g depending on the loss factor of the heated material (Q m ) and the filling factor of the resonator (FF), so that Q u =Q m /FF. When the source is connected through a coupling iris (with a specific coupling factor k), the source conductance G s can be considered transformed to the resonator as G s =k G s. The condition for perfect matching (or in other words, critical loading) of the resonator is G=Gs. With G s <G the resonator is under-loaded and with G s >G it is over-loaded. Typically the loss factor of the heated material changes with temperature and thus the value of G changes. Since G s is constant, we need to modify the coupling factor k in order to maintain good matching and thus optimum energy efficiency of the process. The most popular approach to adjust matching conditions is by the use of a three-stub tuner. There exist tuners with a digitally controlled movement [4][5] but their application is quite complicated. The complication arises from the fact that proper matching is a non-monotonous function of the position of each of the three stubs. Thus it is quite difficult to predict a desired direction of the movement of each stub. It may happen that the movement of one stub produces a local minimum of the reflection coefficient, while the position producing that local minimum is not the one required for obtaining the global minimum. Another problem of a three-stub tuner is possible arcing between metal parts at high power and possible radiation (as considered for example in [6] and [7]). The application of a three-stub tuner could be avoided (or limited) by application of variable coupling. Here, one possibility is a rotating window as described in [8]. Note that the problem of variable coupling to a resonator is also important in particle accelerator technology. This is exemplified by publication [9] describing a complicated setup used for coupling modification. In this paper we will present an alternative solution. It will provide a system for continuous adjustment of the coupling between a mono-mode cavity and an input waveguide, so that good matching will be maintained when the loss factor of the heated material changes. Our solution will consist in the movement of only one tuning element. This will provide monotonous influence on the reflection coefficient measured at the cavity input. The setup will support extremely high power levels and introduce negligible losses into the heating process. Note that variation of the coupling will exert no practical influence on the resonant frequency of the cavity. This is important since microwave power processes often require a correction of the cavity resonant frequency during the heating. When the cavity is equipped with one frequency tuning element and one coupling tuning element, the functioning of the two tuning elements will be practically independent. Such 978-1-5090-2214-4/16/$31.00 2016 IEEE
an arrangement is most desired for simple and fast control of the entire heating process and its best energy efficiency. It is believed that it can be also applied in accelerator technology. II. THE NEW CONCEPT The new concept is introduced to alleviate the aforementioned disadvantages of the three-stub tuner setup by replacing it with a single dielectric-rod tuner. Consider the simplified model of the applicator as presented in Fig. 2. The picture captured from QuickWave-3D QW-Editor [10] presents a 3D view of the model. Fig. 3 shows a schematic view of the horizontal section of the structure. In those two pictures metal parts are invisible. On the left hand side we have a cuboidal cavity with the heated material (magenta) inserted through a quartz tube (green). The cavity is fed by the fundamental mode in WR 284 waveguide shown on the right hand side. The wave is coupled to the cavity through a neck 30 mm wide and 11mm long. Here we shall focus on TM z 110 mode excited in the cavity, with the z-axis indicated in blue in the bottom right corner of Fig. 2. Working with TM z 010 mode in a cylindrical cavity or TM z 210 in a rectangular cavity would not change substantially the subsequent discussion. In the calculated example the quartz tube will be assumed to have its outer/inner diameters of 19.2 mm and 14 mm, respectively. The heated material will have ε =2.8 and the ε changing in a wide range, so that Q m = ε /ε will vary from 5 to 200. Near the neck we place an opening for introducing a dielectric rod. The rod (red in Fig. 2) is supposed to be made of alumina ceramic (ε =10) with diameter equal to about 14 mm placed in a Teflon tube (yellow) with external diameter of 18 mm. Teflon serves as bearing for the ceramic. The alumina rod is ended with a conical part to allow smoother tuning. The diameter of 18 mm is small enough for the wave to attenuate quickly in the opening. According to our calculations, 60 db attenuation can be achieved in about 30 mm of an air tube or in about 50 mm of the alumina-filled tube. We run the first simulation assuming that the alumina rod is moved up and thus remains fully out of the waveguide. Fig. 4 and Fig. 5 present the reflection coefficient calculated with two values of the heated material losses such that Q m =200 (blue) and Q m =5 (red). Fig. 2. A view of the discussed applicator with tunable coupling. Fig. 3. Top view of the structure of Fig. 2. Fig. 4. S11 versus frequency simulated for two considered values of material losses (red for Q m =5 and blue for Q m =200) with alumina rod in the upper position.
tuning range. Thus the position of the tuner must be subject to optimization supported by accurate electromagnetic modeling. The results of the next series of simulations are presented in Fig. 9. The applicator is charged consecutively with materials of Q m1 =5, Q m2 =10, Q m3 =35, Q m4 =95, and Q m5 =200. In each case the alumina rod is moved to achieve perfect matching at some resonant frequency of the structure composed of the applicator and the coupling tuning element. In Fig. 9 five curves of S 11 versus frequency are shown in different colors: Q m1 in red, Q m2 in magenta, Q m3 in brown, Q m4 in green, and Q m5 in blue. Naturally, the width of the matching band depends on the value of Q m. Tuning of the coupling for different losses only slightly modifies the resonant frequency of the structure. During that tuning a maximum recorded deviation of the resonant frequency is only about 20 MHz. Usually such a deviation can be corrected by an easy adjustment of the resonant frequency of the applicator. Thus in practice matching the applicator with an arbitrary material would consist of two steps: first tuning the coupling (by a single shift of the rod that changes the coupling in a monotonous way) and then a minor correction of the resonant frequency of the cavity. The proposed operation is much faster and much simpler in computer control than a matching performed with a three-stub tuner. Fig. 5. Results of simulations as in Fig. 4 presented on the Smith chart. Fig. 4 shows that the assumed upper position of the alumina rod makes the applicator perfectly matched at 2.4483 GHz with the heated material of Q m =200. With the heated material of Q m =5 we have a mismatch with VSWR=40. The Smith chart of Fig. 5 illustrates that in the latter case the applicator is strongly under-coupled. In the next simulation we assume the alumina rod penetrating completely through the waveguide (as shown in Fig. 6). Fig. 7 and Fig. 8 present the reflection coefficient simulated with two values of the heated material losses such that, again, Q m =200 (blue) or Q m =5 (red). Unlike the previous case, we are now perfectly matched for the lossy medium and get a high VSWR (about 40) for the low-loss medium of Q m =200. The Smith chart of Fig. 8 implies that in the latter case the resonator is strongly over-coupled. Note that the movement of the dielectric stub practically did not change the resonant frequency of the cavity. The matching with the lower position of the rod is at 2.4462 GHz thus, only 2.1 MHz down from the case with the upper position of the rod. Our device ensures this very valuable feature only for a specific placement of the tuner, outside the resonator but very close to the coupling iris. In the iris area the wave is evanescent. By introducing a high-permittivity dielectric there we lower the cut-off frequency of that part of the waveguide and thus the coupling through the iris increases. Placing the tuner closer to the resonator would lead to significant changes of the resonant frequency when the dielectric rod is inserted. Placing the tuner further away from the iris would reduce the Fig. 6. A view of the setup with alumina rod in its lower position.
Fig. 9. Results of simulations performed with five different assumed values of material properties, represented by the values of Qm equal to 5 (red), 10 (magenta), 35 (brown), 95 (green), and 200 (blue). In each case the alumina rod has been moved to the position providing perfect matching for one frequency point. Fig. 7. S11 versus frequency simulated for two considered values of material losses (red for Qm=5 and blue for Qm=200) with alumina rod in its lower position. Fig. 8. Results of simulations as in Fig.4 presented on the Smith chart. III. CONCLUSIONS AND PROSPECTS FOR APPLICATIONS To evaluate tuner losses and its power handling capabilities, we have run a further series of simulations assuming: input power of 10 kw, metal conductivity of 3 10 7 [S/m], Q m of alumina equal to 5000, no losses in Teflon, alumina rod in its central position, sinusoidal excitation at the frequency of good matching (f=2.4358 GHz). Under the above conditions we have calculated: power dissipated in metal walls: 170 W, power dissipated in the alumina rod: 6.4 W, maximum E-field in the region of the alumina rod: 330 V/mm. Therefrom we conclude that the proposed tuning device will not require cooling, even at very high power levels. It will also provide many other advantages with respect to previously known solutions: It will be much easier and faster to operate, especially with computer-controlled tuning. Thus it can be applied in industrial processes (like in microwave chemistry), where so far the three-stub solution has been unacceptable (or hardly acceptable). It will be cheaper since it comprises only one simple coupling tuner with no cooling. There will be no problems of parasitic radiation from it. In the example presented herein the loss factor change as 1:40 has been considered. This does not seem to be a limit and we could imagine even a ratio of 1:100, if such a need arises. The example presented in this paper has demonstrated our tuner application for a mono-mode cavity at around 2.45 GHz. The discussed solution may exhibit even more advantages in the case of industrial processes at 915 MHz and power levels of 100 kw or more, where incorporation of any three-stub tuner may be limited by arcing. We also envisage extensions of our concept to the range of other areas of applications with multimode cavities, tunnel ovens, and particle accelerators driven by microwave signals. REFERENCES [1] A. Metaxas and R. Meredith, Industrial Microwave Heating, Peter Perigrinus, London, U.K., 1983. [2] D.C. Dibben, Numerical and Experimental Modeling of Microwave Applicators, Ph.D. Disertation, Cambridge University, 1995. [3] Tunable Vacuum Applicator online: http://www.2450mhz.com/pdf/doc/900081$.pdf
[4] Precision 3-Stub Tuner online: http://www.2450mhz.com/pdf/brochure/ga1001-3brochure.pdf [5] Waveguide Components: 3-Stub Tuner online: http://www.muegge.de/en/products/waveguide-components/matchingcomponent-stub-tuner/ [6] V.Bilik, P.Zajaczkowski, J.Bezek, and W.Gwarek, Electromagnetic simulation and measurement of a tuning stub in R-9 waveguide, MIKON.2004.1358494 IEEE Conference Publications, 2004. [7] V.Bilik and J.Bezek, High Power Limits of Waveguide Stub Tuners, Journ. of Micr. Power and Electromagnetic Energy, vol. 44 (4), pp. 178-186, 2010. [8] R. Kumar, P. Singh, M. S. Bhatia, and G. Kumar, Analytical method for coupling calculations of rotated iris coupled resonator cavity, Progress in Electrom. Research B, vol. 44, pp. 223-239, 2012. [9] R. Wegner, F. Gerigk, J-M. Giguet, and P. Ugena Tirado, Tuneradjustable waveguide coupler (TaCo), Internal note by CERN: CERN- ATS-Note-2011-085 TECH, online: http://cds.cern.ch/record/1387540/files/ [10] QuickWave-3D Electromagnetic Simulator v.2016, QWED, Poland, (www.qwed.eu)