Different polyhedra require different strength magnets

[blaisepascal]

// Pretty Images snipped.

I've seen illustrations like this before, as well as other designs which effectively have one loop serpentining around the entire polywell.

They've always bothered me.

Polywells with traditional coils (like WB-6) use lots of turns to get high B fields with (relatively) low currents. Field strength is proportional to iN, where i is the current and N is the number of turns. So to have a high B, you can have a high i or a high N, or a medium i and N.

For 8" thick coils (ignoring cooling and other effects), to get 1T you are going to need about 160kA-turns. In all of these serpentine designs, I see one turn, which would require 160kA (per T of desired field strength. For a 10T system, that would work out to 1.6MA)

Do the advantages (which have not been specified, as far as I know) of a serpentine design outweigh the engineering challenges of developing a 160kA/T high current power supply?[/i]

[tombo]

The 8" OD of the coil includes layers of cooling and insulation.
The actual SC cross section is more like 3-4" IIRC from a previous discussion of coil sizing.
They do not have just one SC wire.
They have as many SC wires as you would have turns.
It is similar to the MPG.
advantages:
1 Feed lines have full current so have full magnetic shielding.
2 No complicated T junctions between feed lines and coils.
3 No complicated T junctions between feed lines and nubs.
[ Those T's would need to incorporate concentric layers of first wall, hot water cooling layer, insulation, cold water cooling layer, insulation, LN2 cooling layer, insulation, LH2 cooling layer, insulation, LHe cooling layer, superconductor, and throughout it all, mechanical load supports. ]
4 All the complicated stuff is done outside the vacuum chamber where there is a lot more elbow room and repair access.
The SC wires could be connected in series as many-turn coils, or with multiple independent power supplies.
5 Coolant flows are straight forward, if bendy. (one design has only two 45 deg bends and two 120 deg bends, it also has considerable inherent mechanical strength.)


6 Nubs need only transmit mechanical loads.
7 For the simple water or LN2 cooled copper tube in small sizes the manufacture is very simple.
See a discussion called "Magrid Brainstorming" about 1.5 years ago or so for further advantages and further discussion.

Kiteman: The pic above shows the twist that blocks the large angular velocity space loss mode. (Per or earlier discussion. Did that make sense?)

[tombo]

Nubs:
I would sure like to know if nubs on the machine at EMC2 are evenly hot all the way around their circumferences or if they have hot spots. (say toward the north and south poles)
If most of the heat load is on 2 opposite sides then they could be narrower facing that direction and longer the other way to carry the same mechanical load bearing cross section and conductor cross section.

[KitemanSA]

WOW!!!

INDEED!!!! Another thing of beauty from the desk of tombo.

That was a mind bending experience!!!
3 meter dia magrid
8" thick "coils"

IIRC, the ~28cm was the side of the SC box; the casing was quite a bit bigger. But that was before the revelation that the alphas would exit at the cusps.

More views of this?
Note that 60 deg bends are too sharp for normal piping practice but if larger they cause really big gaps between the adjacent bends.

Probably wouldn't be a pipe but a set of forgings/stampings. It would be easy to stamp those parts out of stainless steel. Welding them together would be delicate work.

[KitemanSA]

I've seen illustrations like this before, as well as other designs which effectively have one loop serpentining around the entire polywell.
They've always bothered me.
Polywells with traditional coils (like WB-6) use lots of turns to get high B fields with (relatively) low currents. Field strength is proportional to iN, where i is the current and N is the number of turns. So to have a high B, you can have a high i or a high N, or a medium i and N.

Blaise,
I believe that the standard MPG Icosidodec would look more like this.

Instead of one loop then out, the SC would have one strand coming in and going out thru thinner I/O pipes, but would loop around past the I/O point numerous times. I believe tombo mentioned this possibility before.
The cooling loop would be once around.
There is a variation on this theme where the MPG is divided into two equal loops rather than one which makes the coolng a bit easier.
I am not sure the Icosidodec version of the MPG is a candidate for the X_Cusp treatment like the cuboct version. I haven't figured out how to make the X-Cusps have equal current on all sides like the cuboct.
Without an X_Cusp, each of those little yellow target points would have to have a nub to keep the thing strong enough to handle the Bfield loads; or you could porcupine it.

[KitemanSA]

Kiteman: The pic above shows the twist that blocks the large angular velocity space loss mode. (Per or earlier discussion. Did that make sense?)

Nope.

[KitemanSA]

Without an X_Cusp, each of those little yellow target points would have to have a nub to keep the thing strong enough to handle the Bfield loads; or you could porcupine it.

If I am not mistaken, somebody has suggested that rather than one nub in the middle of the funnyish cusp, there could be one on either side, out of the electron stream.

No wait, that was me.

[hanelyp]

If I am not mistaken, somebody has suggested that rather than one nub in the middle of the funnyish cusp, there could be one on either side, out of the electron stream.

No wait, that was me.

That sounds like a rendering I posted to the old magrid configuration brainstorming thread.

[KitemanSA]

Move the yellowish nubs closer to the point of closest approach, yet not in the hole, and arc the magnet sides rather than angle them, and you have an almost perfet MaGrid, IMHO.