Soft X-Ray tomography optimization on Tore Supra for real-time applications

Abstract

The Tore Supra tomography system presents the capability of delivering into the real time network the Soft X-Ray (SXR) measurement. In order to fully exploit this capability for physics applications, many efforts have been recently deployed in order to deliver very accurate and absolutely calibrated signals with a good signal to noise ratio1.2, in a wide dynamic range. Indeed the ultimate goal of the exercise is to provide in real-time useful information about the plasma parameters which can be calculated from the SXR measurement. This paper describes in details the methodology we chose for getting the local SXR emissivity profile from the raw line of sight (l.o.s.). integrated signals, using a standard minimum Fisher regularization algorithm. The different phases that we followed in order to improve the reconstruction will be presented: first how we ameliorated the quality of the signal by taking into account corrections due to geometry, different electronic amplification, alignments and so on.. Second the different tests we made to enlighten the difficulty of the problem, leading to a particular choice of the pixel grid size and how we suppressed the different artefacts inherent to the chosen boundary conditions. We found the grid 51*51 as best trade-off between spatial resolution, coherent with the number of l.o.s., and computing time. Third, the limitation of the technique due to geometrical considerations and how we checked the robustness of the reconstruction by using artificial signals. Finally, the transposition of the off line code for a real-time application that generated also some adaptation of the inversion technique and grid (35*35 pixels) in order to reduce the computational time will be presented. Examples of analysis performed on the Tore Supra discharges will be shown and perspectives for real-time applications will be then also discussed.