Fusion Engineering and Design 56 – 57 (2001) 615– 619 www.elsevier.com/locate/fusengdes
The ECRH system of ASDEX Upgrade F. Leuterer a,*, M. Beckmann a, A. Borchegowski c, H. Brinkschulte a, A. Chirkov c, G. Denisov c, L. Empacher b, W. Fo¨rster b, G. Gantenbein b, V. Illin c, W. Kasparek b, K. Kirov a, F. Monaco a, M. Mu¨nich a, L. Popov c, F. Ryter a, P. Schu¨ller b, K. Schwo¨rer b, H. Schu¨tz a a
Max-Planck-Institut fu¨r Plasmaphysik, EURATOM Association D-85740 Garching, Germany b Institut fu¨r Plasmaforschung, Uni6ersita¨t Stuttgart, D-70569 Stuttgart, Germany c GYCOM, Nizhny No6gorod, Russia
Abstract The ECRH system of ASDEX Upgrade is now completed. Four gyrotrons generate a total Gaussian beam power of 2 MW/2 s at 140 GHz. The power is transmitted partly quasioptically and partly via corrugated HE-11 waveguides. The transmission line losses, determined calorimetrically, are about 12%. The four focused beams lead to a very localised power deposition in the plasma and can be steered in both poloidal and toroidal directions. © 2001 Elsevier Science B.V. All rights reserved. Keywords: ECRH system; Gyrotrons; ASDEX Upgrade
1. Introduction
2. The gyrotrons
ASDEX Upgrade is equipped with powerful bulk plasma heating systems, namely 20 MW/10 s of neutral beam heating and 8 MW/10 s of ion cyclotron heating. The electron cyclotron heating system with 140 GHz/2 MW/2 s is therefore not designed for bulk heating, rather for very localised electron heating or current drive. For this purpose we use the second harmonic X-mode, and beams with a narrow focus inside of the plasma under all injection angles.
The power is generated by four diode type gyrotrons (GYCOM, Russia) [1]. They can operate in two regimes. In the nominal regime they deliver a Gaussian beam power of 0.5 MW for 2 s, while in the optional regime (at higher beam current) they deliver 0.7 MW for 1 s. The frequencies are in the range 139.5 –140.1 GHz, which places the resonance in the plasma within a range of 9 3.5 mm. Two gyrotrons are fed in common by one power supply via one series tube modulator, and therefore they have the same timing and modulation. In our installation the gyrotrons are situated between 12 and 18 m from the tokamak axis, with
* Corresponding author. Tel.: + 49-89-3299-2223; fax: + 49-89-3299-2558. E-mail address:
[email protected] (F. Leuterer).
0920-3796/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 0 - 3 7 9 6 ( 0 1 ) 0 0 2 7 8 - 2
616
F. Leuterer et al. / Fusion Engineering and Design 56–57 (2001) 615–619
a distance of 2.8 and 4.1 m from each other. The magnetic fields originating from the air core tokamak ( 525 Gauss) and also from the adjacent cryomagnets (54 Gauss) influence the gyrotron performance as described in an earlier paper [2]. Depending on the beam voltage the output power can be varied continously in the range 5–100%. We apply two schemes for modulating the output power. By switching the high voltage on and off we can modulate with a frequency of up to 1 kHz. This scheme is applied for electron heat wave studies [3]. Higher modulation frequency, up to 30 kHz, is achieved by merely reducing the beam voltage to such a level that the oscillation just does not stop. However, since in this case the beam current continues to flow this leads to a high loading of the modulator tetrode. This scheme of modulation was applied in experiments on synchronised stabilisation of neoclassical tearing modes [4].
3. Transmission lines For each gyrotron there is one transmission line to the tokamak. A schematic of such a line is shown in Fig. 1, a description of the components used can be found in [5]. The quasioptical section includes a beam tunnel of : 0.4 m length at the gyrotron output, a matching optics unit (MOU) with two non-quadratic phase correcting mirrors to optimise the Gaussian beam output, and two polariser mirrors. One phase correcting mirror is
Fig. 1. Transmission line schematic, gyrotrons 3 and 4.
equipped with a directional coupler to monitor the forward power. An independent mirror can be switched into the line to direct the beam into a permanently installed dummy load for 2 s pulses. This can be replaced by a water calorimeter load for 250 ms pulses. The Gaussian beam is then coupled into a corrugated HE-11 waveguide line with 89 mm diameter. At the end of these lines a mode transformer section generates a mixture of HE-11 and HE-12 modes, such that their interference leads to a rectangular like beam profile at the torus window, while in the plasma it results in a very narrow beam profile. A directional coupler in the last mitre bend allows to monitor the transmitted power. A switchable plane mirror in front of the window directs the beam either to the plasma or into a calorimeter load. This allows a routine check of the transmission line without the need for a plasma. The transmission lines are operating under normal air pressure. For the adjustment of the mirrors we use liquid crystal foils on an absorbing substrate, or thermopaper. A number of arc detectors survey the whole line to locate possible breakdowns. However, when the line is clean we do not have any arcing problems. Inside the torus a fixed and a steerable mirror (metalised graphite) allow to scan the beam by 9 30° in poloidal and toroidal directions for off axis deposition and current drive. We have studied the power transmission by determining calorimetrically the power at the transmission line input, PTLin, and output, PTLout, and also the power lost in the gyrotron window, which together with a factory determined calibration factor, gives us the total power leaving the gyrotron, Ptot, which includes all non-Gaussian content. These measurements are taken at different beam voltages and in different campaigns. The thermal losses in the load are approximately taken into account. As a reference for comparing the data we take the power measured at the mirror directional coupler in the MOU box. An example is shown in Fig. 2. From a linear fit to the data we then determine the power ratios. The result for all four transmission lines is given in Table 1. The average loss of the transmission lines is 12%, compatible with estimates of the Ohmic losses, beam waveguide coupling losses and mode
F. Leuterer et al. / Fusion Engineering and Design 56–57 (2001) 615–619
Fig. 2. Calorimetric power measurements at different points along transmission line 1.
conversion losses [2], if we consider the latter ones also as lost power. For the boron nitride torus window and the two mirrors inside the torus we estimate an additional loss of 6%. From the gyrotron output to the transmission line input we loose :18% with respect to Ptot. Part of this loss, :7% of Ptot, occurs in the beam tunnel and is due to diffuse radiation leaving the gyrotron at steep angles. The rest is due to beam content which cannot be transformed in a Gaussian beam.
4. Calibration of the polariser mirrors For oblique incidence the X-mode must be excited at the plasma edge with proper elliptical polarisation and spatial orientation. This is done with the two polariser mirrors with groove depths of u/8 and u/4 in the MOU box. They were calibrated with short gyrotron pulses measuring
617
Fig. 3. Calculated detector response as function of the u/4 polariser setting.
the power seen by the polarisation sensitive directional coupler at the end of the HE-11 transmission lines as a function of the polariser mirror positions. The whole set of data was fitted to a theoretical model which takes the non-perpendicular incidence of the beam on the mirrors into account [8], and with the groove depths in both mirrors as fit parameters. Deviations even within machining tolerances introduce noticeable phaseshifts. In Fig. 3 we plot the calculated directional coupler response for an ideal case (perpendicular incidence and exact groove depths of u/8, respectively u/4), for the case with incidence angles of 15° and exact groove depths, and for the case with incidence angles of 15° and the groove depth of the u/8 mirror reduced by 8% (: 20 mm). In this example the setting of the u/8 mirror is chosen to produce a perfect circularly polarised wave in the ideal case.
Table 1 Power ratios determined along four transmission lines
PTlin/Ptot PTLout/PTlin PPlasma/PTlin PPlasma/Ptot
Gyrotron 1
Gyrotron 2
Gyrotron 3
Gyrotron 4
Average
0.81 0.83 0.78 0.63
0.83 0.93 0.88 0.74
0.79 0.88 0.83 0.66
0.85 0.88 0.84 0.71
0.82 0.88 0.83 0.68
618
F. Leuterer et al. / Fusion Engineering and Design 56–57 (2001) 615–619
5. Adjustment of the phase correcting mirrors The gyrotron output beam has a special power distribution to reduce the thermal loading of the BN window and differs considerably from a fundamental Gaussian beam, which is best suited for low loss transmission. Measured power distributions at different locations along the beam were used to calculate the phase distribution for the design of two non-quadratic phase correcting mirrors which optimise the conversion of the complex phase and amplitude structure to that of a Gaussian beam [6,7]. The calculations show that : 85% of the beam incident on the first mirror can be converted into a Gaussian beam. This is compatible with the experimental losses of 18% in the MOU box. To adjust these mirrors we compare the beam pattern on the mirrors, measured with a liquid crystal foil, with the calculated beam pattern. The accuracy of this adjustment is estimated to be 92 mm in a direction perpendicular to the beam. We studied numerically the sensitivity of the resulting Gaussian beam to such alignment errors. The calculation was performed for a given set of mirrors by calculating the propagation of the wanted Gaussian beam backward toward the gyrotron, thus determining an ‘artificial gyrotron beam’. A new forward calculation resulted in a transformation of this artificial beam into a 99.9% Gaussian beam. By changing the mirror positions
we found that differences in the spacing between the phase correcting mirrors of 9 20 mm are negligeable, but a displacement of a few millimetres perpendicular to the beam is of importance. For example, a transversal displacement of the second mirror by 4 mm leads to a reduction of the Gaussian beam content to 96.0%, while such a displacement of the first mirror by 4 mm leads to a reduction to 91.6%. Fig. 4 gives an impression of the distorted intensity contour lines for this example. We note that a few percent of nonGaussian content interfering with the main beam, may lead to a noticeable deformation of the intensity contour lines.
6. Summary An ECRH system with 140 GHz/2 MW/2 s has been installed at ASDEX Upgrade and is routinely in use. The combined quasioptical and HE11 waveguide transmission lines have average losses of : 12% for an incident Gaussian beam. Considerable additional losses occur due to the non-Gaussian beam content of the gyrotron output and due to the losses in the torus window. For both losses considerable improvements can be expected when diamond windows will be available [9]. Special care was given to the calibration of the polariser mirrors including the effects of machining tolerances. Misalignment of the phase correc-
Fig. 4. Calculated intensity contour lines (mm) after beam transformation by two phase correcting mirrors. Left: M1 aligned, Dx=0. Right: M1 horizontally displaced by Dx=4 mm.
F. Leuterer et al. / Fusion Engineering and Design 56–57 (2001) 615–619
tiong mirrors of a few millimeters leads to a noticeable reduction of the final Gaussian beam content.
References [1] M. Agapova et al., Proc. 19th Int. Conf. on Infrared and Millimetre Waves, JSAP Catalog No.: AP 41228, Sendai, Japan 1994, pp. 79 –80.
619
[2] F. Leuterer, et al., Fus. Eng. Des. 53 (2001) 485. [3] F. Ryter et al., Europhysics Conference Abstracts, Vol. 23J, 1999, paper P 4.018. [4] H. Zohm, et al., Nucl. Fus. 39 (1999) 577. [5] M. Thumm, W. Kasparek, Fus. Eng. Des. 26 (1995) 291. [6] L. Empacher, W. Kasparek, IEEE-AP-49 (2001) 483. [7] A.V. Chirkov, G.G. Denisov, Int. J. Infrared Millimetre Waves 21 (2000) 83. [8] Yon-Lin Kok, N. Gallagher, J. Opt. Soc. Am. A 5 (1988) 65. [9] M. Thumm, et al., Proc. Fus. Eng. Des. 53 (2001) 517.
