>>40514176
The ideal cavity geometry would likely be saucer-like or toroidal, but the best geometry would likely be obtained by finite element analysis and computational design methods. The RF frequencies here would be, at least, in the GHz or THz range. In a highly conductive cavity with a high Q-factor and a conductor-dielectric interface, you would have plasmon oscillations in the cavity wall. If someone built a working saucer back in the 1940s, like in the Haunebu legend, they would have needed a very high-frequency magnetron or klystron source of high quality. Biasing the field and steering would be very tricky. Many early manmade saucers (such as in the Haunebu legend), are rumored to have a "fast-rotating mercury plasma", such as in the so-called "Thule Triebwerk". I hypothesize that this device was indeed real, but its mode of action was improperly explained. The pressurized, spinning mercury plasma would produce intense, rotating B-fields. There's your magnon source. Couple them together and you have a magnon-plasmon polariton and, ultimately, a hopfion with its own winding number.
>each plasmon carries angular momentum
>collective plasmon state has net helicity
>at critical density, helicity forces topological winding
>the winding number is the hopfion charge
You are basically stuffing an MRI magnet into the guts of a closed radar horn.