I like the design, if I understand it correctly:
I take it you have:
- An outer ring of metal,
- welded-on mounting lugs,
- a strip of metal (or set of rods) between each pair of coils, edge-on to the magnets and welded on to the outer ring, providing support and separation for the coils.
- An inner ring of metal with the other end of the strips or rods welded to it.
- Coil hold-down clamps composed of rings of metal outside the pole path, screwed to additional mounting lugs.
Yes, such a setup would have relatively low eddy current issues. The stuff outside the magnet pole path isn't subject to much field and the stuff within the pole path is thin and edge-on to the field path.
The main current loops would be through:
- the outer ring,
- the radial strips/wires,
- the inner ring, and
- back through the radial strips/wires.
With an additional similar path through the clamping structures. Yes, this is not all that great a loss, especially with reasonably thin and resistive material.
But you could do still better:
If the inner portions of the radial structure are insulated from each other you break the loops.
Alternatively you could take advantage of the fact that the induced voltage along every third radial strip is the same. So for two out of three radial strips break them in the middle, or somewhere between the inner and outer clamps (unless the set of clamps on the "break" end are held on only to every third strip, or held with non-conductive screws or insulated by shoulder washers on the ohters.) Bingo: No currents around the loops. (You can fasten the cut ends together with a non-conductive epoxy to avoid abrasion if the break is where the coils would rub on it.)
Another approach would be to have THREE inner rings, insulated from each other, with each ring connected to 1/3 of the radial structures. And another approach to the inner clamp rings would be to similarly have three of 'em, each screwed to 1/3 of the radial structures.
With this arrangement eddy current losses are comparable to that in a copper wiring turn of comparable thickness, i.e. nearly non-existent.