A method to determine arc position in ICRF systems

作者:D' Inca R*; Polozhiy K; Eckert B; Siegel G; Noterdaeme J M
来源:Fusion Engineering and Design, 2013, 88(6-8): 895-898.
DOI:10.1016/j.fusengdes.2012.11.025

摘要

The study of arcs on ICRF systems has mainly focused on the development of fast detectors to detect a breakdown, independently from its position, before they can damage the components [1]. However, the ability to localize an arc is also an important factor: arcs can be the sign of a defect and if they occur at the same location, they give a good indication of the position of this defect without having to use intrusive diagnostics like optical fibers. We present here a method to find the position of arcs based on the analysis of their high-frequency noise. Arcs are indeed fast transient which excite frequencies in the MHz range. The SHAD system reacts on the integral value of this noise signal to detect the development of arcs. But a further investigation of this signal shows detailed features like splittings and chirpings of the frequency peaks which give more information on the source of the breakdown: at its birth, the arc emits a short electromagnetic pulse of several nanoseconds that propagates inside the ICRF system; this pulse is reflected by components like the antenna, the stub tuners or the generators back to the arc thereby changing its boundary conditions. This resonance between the arc and the ICRF system can be measured and the time of propagation of the pulse gives useful data on the position of the breakdown. We have carried out tests on the Manipulator eXPeriment (a testbed to artificially create arcs in a representative ICRF system) and used a fast digitizer at 2 Gigasamples/s to capture the signal of breakdowns at different locations. A RF model of the testbed was developed to simulate the theoretical propagation of this signal and to compare it with the measurements. This makes it possible to determine the accuracy of the position measurement and to evaluate the complexity of the model required to get a good estimation of the position. Indeed, to be useful and robust, the modeling should be kept as simple as possible.

  • 出版日期2013-10

全文