Losses in a multi-cell accelerating cavity:: measurement against simulation

In the Trispal CCL accelerating cavities, the coupling slots between contiguous cells had been optimized with MAFIA, in a way to maximize the shunt impedance. The result was a set of “4-petal” slots that were supposed to yield almost no shunt impedance drop, compared to a single cell cavity [P. Balleyguier, Linac conf. Geneva, 1996, p. 414]. In order to check this out, a cold model has been constructed. The goal was to accurately measure the Q-drop caused by coupling slots, and to compare it to predicted values. We achieved a good experimental precision on the Q-drop, and we can state that the predictions were definitely wrong: the actual Q-drop is (−22.5±0.5) %, about four times more than simulated. We conclude that the “4-petal” slots should be abandoned, and that 3D-code estimations of local power loss may be very inaccurate in some cases.