In fusion experiments the magnetic diagnostics provide the signals to calculate the main electrical parameters of the discharge, such as plasma shape and position. These signals are the induced voltage on passive pick-up coils and full or differential flux loop, the desired field and flux are obtained by analogue integrators. The use of Hall effect sensors would directly provide the field values but they cannot withstand high neutron fluxes. The evolution of fusion experiments shows a continuous increase of the plasma pulse duration: in JET plasma duration of the order of a minute was reached and the expected plasma pulse in ITER should be initially about one hour long. Therefore it is mandatory to have the capability of integrating the signals from magnetic probes with very small drifts. The possibility of realizing long pulse analogue integrators in a time scale of several hours was demonstrated [1]. Alternative systems able to provide an output proportional to the field were proposed. A hybrid method was presented based on the measurement of the torque on a coil in which a known current circulates and on the conventional measurement of the induced voltage, for measurement of the field at low and high frequency, respectively [2].

In order to understand the possibilities of applying this method in fusion devices, it was considered useful to test a sensor based on the measurement of a torque and to analyse the accuracy that could be obtained. As a first step, a probe was designed to measure a stationary maximum field of 100mT along a single direction, aiming at an absolute uncertainty of few millitesla. Care was used in the choice of the material to minimize the perturbation on the external field due to the presence of the sensor. The measurement of the torque is obtained from the outputs of two strain gauge cells and the distance between them, in such a way errors in measuring the induction field due to the unavoidable lack of symmetry can be reduced. The paper deals with the results of the sensor calibration in a known magnetic field: offset error, linearity, hysteresis and reproducibility of the measurements are taken into account.

[1] S. Ali Arshad, L. DeKock, Long-pulse analog integration, Rev. Sci. Instrum. 64 (9), September 1993
[2] R.D. Woolley, Tokamak poloidal magnetic field measurements, accurate for unlimited time durations, Proc. of Symp. on Fusion Energy, 1995