icarus wrote:Reading your comments here I think you, and others, may have missed the point of this exercise.
The field goes the same direction for all the faces (out or in) and the opposite direction out (or in) in the small line cusps gaps between the coils.
It is an identical mag-field topology to WB-6 (7, 8?) but the field coming through the line cusps have been squeezed to their minimum (order of a gyroradius length scale). I.e. the smallest gap between the circular coils of WB-6,7 becomes the gap for the whole length of the line cusps of those machines (which is the same length for this configuration).
The rationale being, if the line cusps are the problem squeeze them up as much as possible. Obviously, they cannot be sealed shut completely or that would be a different field topology altogether.
What is clear from the morphology of the mag-field is that by a simplified conservation of flux-tubes the same flux through a central-point cusp is to be expected issuing from its attendant (surrounding) line cusp. (hat tip to Art Carlson on this one)
An analogy would be if you squirt a jet of water at a surface on the perpendicular. The mass flux of water through the diameter of the jet (central cusp) is the same mass flux of the water through the thin annular ring exiting the impact area (line cusp), regardless of the diameter of surrounding ring (length) you choose to analyse.
The first part of your comments make sense to me, it seems similar to what I argued above. But, the last part does not make sense in light of claimed efficiencies. Perhaps you point is valid in an imaginary system.
If a ring magnet was made of of an imaginary thin crossection , then the outer line cusp might have a surface area similar to the central point cusp, because the point cusp has significantly weaker opposing magnetic fields, while the line cusp is much longer. This, along with the next magnet being placed very close to the reference magnet (the line cusp is much longer, but also much narrower than the point cusp), and no deviation in the distance between the magnets (no curving of the magnets, ie: square coils with sharp 90 degree corners. I can see this imaginary system having equal point cusp and line cusp areas. This picture does reinforce my current belief that the corner areas of the magrid dominates the line cusp losses. Otherwise, higher order polyhedra (which introduce increasing line cusp lengths (I think)) would leak more if the corners did not dominate losses (I think the corner areas would decrease). If this is real and significant, it would improve confinement. Or, the anticipated gains might be completely due to increased quasi sphericity.
But, in the real world (especially when the need for several gyroradii spacing of the magnets was realized) the situation is much different.
The spacing increases the effective width of the line cusps. And the crossection of the coils is not infinitely small. The minor radius of the WB6 coils was ~ 17% of the total diameter. This effectively moves the maximum B-field area towards the center of the coil (point cusp) and away from the line cusps. As the B field strength drops with the square of the distance, this effect is considerable. In a super conductor where there may only be a few thin windings, this effect may be less pronounced(?).
I have wondered if making the coils thicker- especially in the direction towards the center (oval shape instead of circular crossections) would help much. As the point cusps are already considered to be a minor loss area, the benefits may be small. Other concerns about crossectional area exposed to cross field transport, x-rays, neutrons, etc may cause more problems than any minor gains.
[EDIT] The original 4 grid spherical layout that started this thread. would suffer from longer point / face cusps. I assumed it would not be a severe penalty compared to the anticipated gains from reducing the corner cusps. One thing that I have considered is varying the crossection shape of the magnets to mitigate this somewhat. The internal volume of the coil would not change, but the shape would. Near the poles, the coil would be thicker in the direction towards the point cusp, while the coil would be thicker away from the center of the Wiffleball in the regions near the equator of the coil ( the two corner cusps are considered as the poles). This would presumably decrease the point cusp area, perhaps increase the corner cusp area, but the balance may be beneficial, again depending on the relative leakage rate through the cusps, along with other possible concerns, and optimizations.
Dan Tibbets
To error is human... and I'm very human.