Estimation of the Loss in the ECH Transmission Lines for ITER S. T. Han, M. A. Shapiro, J. R. Sirigiri, D. Tax, R. J. Temkin and P. P. Woskov MIT Plasma Science and Fusion Center, MIT Building NW16-186, 167 Albany Street, Cambridge, MA 02139, USA Email: temkin@mit.edu D. A. Rasmussen, Oak Ridge National Laboratory, PO Box 2008, Oak Ridge, TN, 37831 Abstract - The transmission lines (TLs) for ITER electron cyclotron heating (ECH) require extremely low losses. The maximum loss that can be tolerated is close to the theoretical value estimated by the ITER team. We are investigating methods to precisely estimate the extremely low losses in the TL components for ITER. We present calculations of the loss in the ITER ECH transmission lines, including both intrinsic and extrinsic losses of all of the components, and discuss possible approaches to reduce the losses. Low power (cold test) experimental results on the measurement of the loss in some TL systems with coherent and incoherent techniques are presented and compared. 1. Losses in ITER TL TABLE I: LOSS FOR ITER TL. Losses ITER DDD 5.2 This estimate MOU loss 0.22 db <0.01 db Injection loss Coupling loss, Tilt, Offset 0.035 db 0.116 db Intrinsic loss Miter bends (7) 0.248 db 0.19 db Polarizers (2) 0.044 db 0.066 db Extrinsic loss 100 m WG WG Sag 0.078 db 0.039 db WG Tilt/Offset 0.036 db Other loss (many items) 0.025 db 0.043 db Total loss 0.65 db / 14% 0.49 db / 11% Loss without MOU Loss 0.43 db / 10% 0.49 db / 11% The ITER ECH system includes 24 170-GHz gyrotrons and 3 126.7-GHz gyrotrons. The RF power is transmitted using circular corrugated waveguides with a diameter of 63.5 mm. According to the ITER Detailed Design Document Section 5.2 (DDD 5.2), the transmission loss is limited to 14 % in the transmission lines themselves and 3 to 4% in each of the four ECH launchers (3 Upper launchers and 1 Equatorial launcher). The total loss should not exceed 17 %. The power of 20 MW generated by the 24 1-MW gyrotrons at 170 GHz, should therefore be delivered to the plasma with a transmission efficiency of 83 %. Table 1 contains
our estimation of losses in the ITER T/L and a comparison to the results from the ITER Team published in DDD 5.2. The transmission losses estimated in DDD 5.2 include the stray radiation losses in the Mirror Optics Unit (MOU), or the RF conditioning unit (RFCU), connecting the gyrotron to the T/L entrance. The MOU losses are estimated to be 5 % according to [1]. In contrast, we assume that the requirement for the 170 GHz gyrotron power is 1 MW in a Gaussian beam (TEM 00 mode). This requirement is taken into account in our loss estimation. The coupling of a Gaussian beam to the HE 11 mode of a corrugated waveguide is 98 % for the optimum round beam waist and no tilt/offset. Our estimation includes tilt, offset errors, and beam waist deviations. The injection loss due to coupling the TEM 00 mode into the corrugated waveguide mode HE 11 is 0.116 db, as listed in Table 1. The intrinsic loss in T/L includes the loss in the miter bends and other components, as well as the Ohmic losses. The mode conversion losses in the miter bend are calculated using the gap theory [2]. We calculated the mode conversion loss as P mc =0.011 db. The Ohmic losses per miter bend (averaged for H- and E-plane polarizations) are P ohm =0.006 db. The miter bend losses due to tilt/offset are 0.01 db. Therefore, the losses per miter bend are 0.027 db. For 7 miter bends, this results in a loss of 0.19 db (Table 1). The losses in a pair of miter bend polarizers are calculated separately. Taking an average of double the losses at a flat mirror, we get P ohm =0.008 db for the H-plane polarizing mirror and P ohm =0.016 db for the E-plane polarizing mirror. Such polarizing mirror includes mode conversion losses of 0.011 db and manufacturing error losses of 0.01 db. Added with the ohmic losses, this yields a total loss of 0.066 db for the pair of polarizing mirrors (Table 1). The extrinsic loss in T/L includes the loss due to sag and tilt/offset of waveguide sections. These losses might be significant if the T/L is not built or assembled properly. Mode conversion occurs due to the waveguide sagging, tilt at the waveguide junction, and offset of the waveguide axes at junctions. In this estimation, we follow [3]. We use a waveguide section length of 2 m and assume that the sag curvature radius is <1000 m. The estimated mode conversion loss in a 100 m long T/L is 0.039 db (Table 1). For the assumed tilt angle of 0.001 rad., and the offset of 0.1 mm, the tilt/offset losses are 0.036 db (Table 1). The total estimated loss in the T/L is 0.49 db (11%). Therefore, our estimate is close to the ITER Team estimate of 14 %. The ITER requirement for the T/L loss can be met. It may be possible to reduce the intrinsic loss with improved miter bends.
2. Waveguide Gap and Miter Bend Simulations Since some of the losses are estimated based on current theory for waveguide gaps and miter bends, it is important to verify the validity and accuracy of the theory. To do so, we ran a series of simulations using Ansoft s HFSS at different values of the ratio of waveguide radius to wavelength (a/λ). These first simulations were for a waveguide gap and were carried out for the TE 01 mode in smooth wall circular waveguide. The HFSS simulation used a 4º section of waveguide split by a gap of length 2a. Due to symmetry, only TE 0n modes are considered; however, a larger simulation confirmed that conversion to other modes was negligible. The results for transmitted power are shown below, and are plotted against the approximate TE 01 gap theory from [2]. 1 Transmitted TE01 Power 0.9 0.8 0.7 0.6 0.5 0.4 Theory Simulation 0 2 4 6 8 10 12 a/λ Fig 1. Comparison between TE 01 gap theory and HFSS simulation While there are some noticeable differences between the simulation and theoretical results, the gap theory is within 1% of the simulated values for the high values of a/λ of concern for the ITER TL. The simulations also showed that lower order modes, TE 02 -TE 05, had about 2-3 times the magnitude of higher modes at high values of a/λ, and that mode conversion to a particular mode would peak when closest to its cutoff. The simulations also showed that the lost power primarily appears as radiation passing out through the gap, rather than as high order modes of the output waveguide. Preliminary simulations for the miter bend were also run; however comparisons with more powerful codes such as Star, Inc. s ANALYST are necessary before any conclusions can be reached. 3. TL Loss Measurements
For cold test measurements, we use a vector network analyzer (coherent technique) and a radiometer (incoherent technique). -0.5-0.6-0.7 Loss (db) -0.8-0.9-1 -1.1-1.2 138 139 140 141 142 Frequency (GHz) Fig 2. Losses of a 140 GHz demonstration TL (1.7 m long corrugated waveguide with 12.7 mm dia. and three miter bends) measured by vector network analyzer For proof of principle, we conducted precise measurements of losses in a commercial 140 GHz TL (built by Thomas Keating, UK) with a diameter of 12.7 mm. We used corrugated horns for launching the HE 11 mode from a WR8 rectangular waveguide. Time domain filtering was employed for the data obtained in the frequency domain. This involved creating a gate which includes the first peak and excludes all others (round-trip reflections) by zeroing after the gate in the fast Fourier transform (FFT) of the S21 [4]. To increase the number of data points for the FFT, a discrete sweep over a broad band (100 147 GHz) is applied. Two hollow dielectric waveguides (4 inch long and 0.75 inch thick G10 tube) are inserted between the tapered horns and corrugated waveguides to exclusively damp unwanted higher order modes generated at the mode converters and tapers. The HE 11 mode in the dielectric waveguide attenuators is the same as that in the connecting corrugated waveguide, which has the same inner diameter, provided the loss through the tube wall is large enough for the tube to be considered infinite in radial extent [5]. The loss measured by a vector network analyzer for a system of 3 miter bends and 1.7 m of corrugated waveguide is around 0.9 db (0.3 db per miter bend) at 140 GHz, which is reproducible within ±0.02 db. (Fig 2) To benchmark the cold-test results measured by the vector network analyzer, a 137 GHz heterodyne radiometer was used with broadband thermal radiation from a liquid-nitrogen
cooled 30 mm-thick pyramidal-surfaced eccosorb black body [6]. The radiometer consists of a 137 GHz local oscillator and intermediate frequency (IF) amplifiers covering a 2 GHz range, allowing us to measure from 135 to 139 GHz. A corrugated horn provides an HE 11 mode field-of-view that was coupled by a hollow acrylic plastic conical transition. A chopper permits loss measurements as a difference in the receiver noise temperature with lock-in amplifier phase sensitive detection. The average loss measured by the square-law detector is about 0.25 db per miter bend. This value is within the error bounds of the loss measured by the vector network analyzer. The loss predicted by analytical theory for this demonstration transmission line is 0.17 db per miter bend. The reason for the discrepancy between experiment and theory is not understood at this time. It may possibly derive from errors in fabrication or assembly of the parts. We will soon apply the same techniques to measuring the loss in the 170 GHz ITER TL parts from General Atomics (GA). Those results will then be directly compared with high power test results from the 1 MW, 170 GHz test line at JAEA, Japan. This work is funded by the US Dept. of Energy through the ITER Project Office, Oak Ridge National Laboratory. References [1] A. Kasugai et al., 1 MW and long pulse operation of Gaussian beam output gyrotron with CVD diamond window for fusion devices, Fusion Eng. Design 53 (2001), 399-406. [2] J. L. Doane and C. P. Moeller, HE 11 Miter Bends and Gaps in a Circular Corrugated Waveguide, Int. Journ. Electronics 77 (1994), 489-509. [3] J. L. Doane, Design of circular corrugated waveguide to transmit millimeter waves at ITER, to be published in Fus. Sci. Tech., 2008. [4] M. E. Hines and H. E. Stinehelfer, Time-Domain Oscillographic Microwave Network Analysis Using Frequency-Domain Data, IEEE Trans. MTT 22 (1974), 276-282. [5] E. A. J. Marcatili, and R. A. Schmeltzer, Hollow Metallic and Dielectric Waveguides for Long Distance Optical Transmission and Lasers, The Bell System Technical Journal (1964), 1783-1809. [6] P. Woskov, et al., Corrugated Waveguide and Directional Coupler for 250-GHz Gyrotron DNP Experiments, IEEE Trans. MTT 53 (2005), 1863-1869.