The measurement of neutral beam thermal profiles on 'V'-shaped calorimeters

It is customary in high power neutral beam systems to use a 'V'-shaped calorimeter to stop and measure the beam. With proper instrumentation, it is possible to determine both the neutral beam power and divergence. By utilizing a near-grazing angle of incidence, the area over which the beam is in contact with the surface is increased, thereby decreasing the power density over the case of normal incidence. Thermocouples on the back of the calorimeter, in conjunction with real time fitting algorithms, are used to deduce the divergence from the thermal profile. This measurement implicitly assumes that the measured-profile corresponds to that of the incident beam. It is shown that such is not the case. Energetic particle reflection at near-grazing angle causes the thermal profile on the calorimeter to be more peaked than the incident distribution. The implications of this on the non-linear multiple regression technique of determining the divergence are discussed. With the aid of a reflection model, developed and applied to the beam from a typical TFTR ion source, it is shown that a peaked power density can be modelled. Neural networks are being studied as a means of supplanting the older regression technique of measuring divergence. Y-direction divergences have been successfully derived using a one-dimensional neural network.