electromagnet design

[bwana]

One of the problems of the polywell is that the ions (and electrons) escape through the holes of the wiffle ball. This is a basic design flaw of using 6 magnets arranged on the faces of the cube. One solution might be to invent a nested polywell so that the 'outer polywell' has the faces of its electromagnets covering the truncated corners of the inner polywell. This gets complicated real fast especially since you have the outer magnetic field pushing on the structure of the inner polywell.

An alternative to this steampunk design might be to rethink the idea of the electromagnet. Instead of having a bunch of windings around a circle to give a doughnut, help me understand what might happen if each loop wound was offset from the last one by the diameter of the wire. Instead of coiling the wire around a circle, you would lay it on the surface of a sticky sphere. As the windings are layed down, they would look like a helix that is smeared over the surface of the sphere. As you wind and wind, the equator of the sphere gets covered in the helix. Eventually you come around to where you started. At this point you have an electromagnet whose fundamental design is 2 dimensional (not 1 dimensional like all the electromagnets made) The north pole is really the north 'equator'.

Now continue winding but move to a different latitude. You probably have to keep spraying this thing with enamel spray paint to keep it sticky. Again you get another north equator slightly offset from the last. Eventually, you get the whole sphere covered. Now you dont have a north equator anymore but a north 'shell' facing in. And no holes for the leakage of electrons and ions.

An alternative winding pattern would be to change longitude instead of latitude. This would give a series of north equators symmetrical about an axis.The field would be asymmetric (strongest at the poles of the axis) but probably easier to construct. Also, the poles would make good injection sites-catching and deflecting the ions injecting from the opposite pole without disturbing the symmetry of the magnet shell.

Has this ever been done? My googling has been fruitless.
(Although I really prefer the snazzier search engines like
http://search.spacetime.com/
http://www.search-cube.com/
)

Is there a way to simulate this so I can see what the field looks like? There must be software where the curve can be typed in (the formula for a helix like the one i described is not hard).

Does anyone here know how to make magnets? It seems that a CNC milling machine could be adapted to the laying down of wire -just turn the milling bit off and use a needle in the chuck - upside down. The eye of the needle would be exposed and the chuck would be gripping the pointy end. The wire could be fed through the eye of the needle and the position of the chuck would be computer controlled. An alternative would be to hire a meticulous swiss watch craftsman to lay down the wire.

thanks.
stefan

please don't laugh too hard at my naivete.

[D Tibbets]

I'm uncertain of the patterns you are proposing. I envision something like a choke in electronics. As I understand it these serve to create competing magnetic fields when current passes through the wire and thus act as a impediment to the flow of current/ signal.
In a Polywell the cusps are an essential part of the design. You need these 'ports' for input of electrons and ions. A neutral gas setup might be different from the created ion standpoint, but not the electrons. Recirculation is also a key part of the design. Also, at least for fusion product extraction/ direct conversion you need these exits. Bussard mentioned one configuration that had been studied by others in which many magnets with alternating poles were laid on a sphere. It apparently does not work.

Magnetic strength is additive, the more windings you have going in one direction, the stronger the magnetic field, Having wires winding in various directions would impede this to a large extent. Each time you change directions or displace the path of the wires you are essentially creating a new magnet and the overlapping wires would essentially create very many tiny and weak magnetic cusps. You would lose magnetic strength, adequate separation between the magnets for recirculation, and greatly multiply the corners where magnetic cusps meet the surface of the wires. IE: there would be more metal exposed to the electrons and the magnetic field strength would be greatly decreased.
In some sense this may be like having very many faced polyhedra, except there would be the great disadvantage that the cusps between the wires/ wire bundles would be intercepting the wires almost everywhere.
Perhaps you are thinking that this would form a cuspless spherical structure, but from my reading this is impossible. There would have to be at least two polar cusps, even if you managed somehow to prevent cups between the wires. This is the advantage in containment of a torus shaped confinement- the circle shape brings the two polar cusps together so that there is no effective cusp. This is good for containment except it does not allow an exit port (thus diverters) and it apparently cannot be configured so that there are not regions where the magnetic surfaces are not sometimes concave towards the center. This creates a whole new set of problems with macro instabilities. The cusps in the Polywell avoids this problem because there is no magnetic field (and hopefully these magnetic field absent holes are very small) and thus no covcave fields between the magnets. You are trading these stable holes against torus macro instabilities. Despite this, the containment efficiency of this cusp design is very much less efficient than Tokamak type confinement (perhaps on the order of ~ 10 milliseconds VS up to 1000 seconds. That is a deficiency of ~ 100,000.
But the tradoffs are apparently more than compensatory. The electrons are only contained by this magnetic scheme. The ions may be electrostatically contained longer, or more importantly, they are not lost much faster, as they would be in this sized machine due to only cross field transport.
Due to other considerations such as the negative potential well and the significantly higher ion densities that these machines can maintain for these brief millisecond time frames the fusion rate is fast enough that a significant portion (or even most) of the fuel ions will fuse within this time scale, so there is no need to confine them longer. Combined with other claimed processes this time frame is an advantage as there is not time for the ions to thermalize. This provides additional advantages.
The reason that Tokamaks are so large is because while there are essentially no cusps, the ions have to be contained magnetically. This containment is directly related to the strength of the magnetic field and the rate at which the ions are scattered deeper and deeper into the magnetic field until they impact a solid surface. This distance the ions are scattered for each scattering collision is dependent on their mass. In the Polywell the ions are contained mostly by the electrostatic potential well created by electrons. The electrons also move through the magnetic field in a similar fasion, but the distance they move by this gyroradius giuded scattering is ~ 60 times or more smaller (square root of the mass difference between electrons and protons). Essentially, this allows the linear scale of the magnetic fields to be ~ 60 times smaller for the smae confinement time- this is why Bussard said that magnets are no 'darn' good for containing ions- they can only impede them for awhile.

Another very important aspect of the Polywell cusp fields (no concave magnetic fields towards the center, is that this is claimed to allow for much greater pressure buildup of charged particles within the resultant Wiffleball. This increased density is what allows for the relative poor confinement to be good enough. If the density is ~ 1000 times higher than in Tokamaks, as is claimed, then the fusion rate is that difference squared, thus the same fusion output can be achieved in ~ 1 millionth the time- thus the relevant difference in needed containment times. It also allows for significantly smaller machines to achieve the same fusion output.

Essentially , your suggestions would seem to have two disadvantages to the separate magnet approach. Though your cusps might be much smaller size, and the metal exposed at the edges of each of these tiny cusps would be smaller, the number of cusps would be tremendously larger with the net effect that more surface area is exposed, so recirculation would be worse. Secondly, the magnets would be weaker, thus the cusp would be relatively larger, Wiffleball compression would be restrained, cross field transport (or diffusion) would be greater.
The sphericity advantages are apparently limited. Bussard claimed that approaching a spherical average Wiffleball border might result in a maximum gain of ~ 3-5X. If the smaller conductors in smaller more numerous magnets results in an obtainable magnetic field strength per magnet of EG- 0.1 X of the larger magnets, then with predicted yield scaling at B^4, this would result in a fusion rate of ~5/ 10,000. A definite loss. So the modest gain possible with greater sphericity competes with the volume needed for the insulation and cooling of the wire bundles, irregardless of how efficient the wires are at conducting electricity. Even a ridiculously small and extremely high temperature- say room temperature, superconducting wire that might be able to carry 10 million amps of current would need substantial insulation and cooling to survive the perhaps MegaWatts of heating the wires would be exposed to. The heating would come from x-rays, neutrons, gamma rays, and any fuel or fusion charged particles that managed to reach the magnets.

My impression from the 2008 EMC^2 patent application, is that In WB4 the electron cusp losses were ~ 100 times greater than cross field transport losses. In WB6 the more efficient recirculation may have improved this to ~ 10 X greater cusp losses (I might be off by an order of magnitude- 1000X, and 100X respectively). It was hinted that if recirculation efficiency can be improved further (such as by moving the nubs as might have been done in WB7.1) .the cusp losses may approach the cross field losses. This would reduce the required electron input current and improve performance. But it illustrates the thermal loads the magnets might be exposed to by this direct charged particle ( primarily electron) impacts. If the total input electron power is ~ 1-10 MW with a recirculation efficiency that is ~ 1/10th of the magnatic cross field transport efficiency, then the coils would be exposed to ~ 100,000 to 1,000,000 Watts of direct heating. This ignores the other sources of heating. This is not a trivial heat load, especially if you need to maintain liquid nitrogen, liquid helium, or even room temperature conditions. Significant active cooling will be needed- which means insulating and cooling plumbing will have to surround each wire/ wire bundle.
While your winding pattern theoretically could reach some tiny microscopic cusp demensions where the close proximity of the weaker magnetic fields would overcome the exposed surface ratios, from a practical stand point, it would be impossible to engineer such a condition.

PS: As if I haven't rambled on already. You also have to consider the separation between the wire/ wire bundles/ magnets. You have to have several gyroradii of the electrons separation to allow for recirculation. This may be several millimeters. So much for microscopic separation of adjacent magnets. Also, I should emphasize that these closely and even overlapping wire bundles would not create a cusp less sphere (not even the two cusp theoretical sphere mentioned above).. My understanding is that if you could do that, you would have created a practical monopole, and you would be assured of a Nobel Prize.

Dan Tibbets

[rjaypeters]

I was writing a post but Dan Tibbets beat me to it. Just as well, he writes better than I do.

I invite your attention to the \Theory\magrid configuration brainstorming thread of this forum which starts here:
viewtopic.php?t=289&start=0

Take some time to skim through this thread and your naivete, at which I'm NOT laughing, will be at least better informed. The thread started in late 2007, so if you read it, you'll be telescoping three and a half years of thinking into a short period. A good investment of time, I think.

An alternative to this steampunk design might be to rethink the idea of the electromagnet. Instead of having a bunch of windings around a circle to give a doughnut, help me understand what might happen if each loop wound was offset from the last one by the diameter of the wire. Instead of coiling the wire around a circle, you would lay it on the surface of a sticky sphere. As the windings are layed down, they would look like a helix that is smeared over the surface of the sphere.

I believe you are describing some variety of prolate cycloid (it has loops at the end instead of points) inscribed on the surface of the sphere.

I am currently the most active modeler on this forum and your ideas intrigue me. I'll think about them.

[KitemanSA]

One of the problems of the polywell is that the ions (and electrons) escape through the holes of the wiffle ball.

Two objections with this statement. First, the ions can only escape if they have gotten enough energy to climb to the top of the potential well. Then, the wiffleball should turn most of them back. Second, the electrons don't "escape" the wiffleball unless they too have up-scattered where-upon most of their energy is recouped climbing the external potential well.

So, yes the cusps should be kept small as practical but no, don't go overboard trying to prevent the unpreventable.

[D Tibbets]

Just to be picky, KitemanSA, uses 'escape' here as total escape. This is a two step process. First, all electrons try to travel down their potential well, so if they hit a cusp, they will escape magnetic confinement. The second step, recirculation- captures these magnetic confinement escapees, and reverse their course and accelerates them back into the magrid. Two important points about recirculation (my understanding of it). These escaped electrons may have a kinetic energy of a few to 11,999 eV. If the magrid has a 12,000 positive potential, all of these electrons will be reversed and be accelerated back towards the center of the magrid. There are some inefficiencies so they will not create or sustain a potential well equal to the accelerating voltage, but perhaps ~ 85% of this. This action restores the recirculated electrons to initial monoenergetic KE and also, restores the central convergence of the electrons. The exact same thing could be done with new electron gun provided electrons while the magnetic confinement escapee electrons are lost. The difference is that the recirculation does this at almost no energy cost, and is essential in order to reach breakeven with any fuel (except perhaps D-T?). The exception to this very low cost recirculation is if the magnetic escapee electrons have been up scattered enough that their KE exceeds the voltage on the magrid. These continue on their path to the walls, but still 12.000eV of their energy has been recovered. So these upscattered electrons , which would otherwise tend to thermalize the entire electron population, are removed at a discounted cost. This is very important to help to maintain the non thermal distribution of the electrons within the machine. Both the cusps and recirculation are essential. I suspect that at some point, if the cusps losses are reduced too much, the upscattered electrons would not be removed fast enough, and thermalization penalties would outweigh the input energy gains.
Again, this illustrates the various balancing acts that are claimed to allow for positive energy production. It's sort of like a Goldilocks story.

Note that Cusp confinement with resultant Wiffleball effect PLUS recirculation is what can lead to optimal confinement and delay of electron thermalization, not just trying to minimize cusp confinement losses alone.

Dan Tibbets

[CosmicObserver]

Hello all,

This is my first post, and am very intrigued by the Polywell concept.

I have read through all 38 pages of the brainstorming thread, but have not seen any proposals for different cross-sections of the torus. We know square is bad, but what about an elliptical one, with it's major axis pointed directly at the centre. The donut would fit onto an imaginary cone with the tip at the centre and it's base on the face of the magnet, but be 'squashed' towards the cone's surface. The electrons, ions and fusion products created at the centre would see a much smaller cross-section. As long a the ellipse didn't have too great an aspect ratio, recirculation around the magnet shouldn't be much worse than a circular cross-section.


This would also give a much greater area of interaction between two adjacent magnets. Could this be used to any advantage with respect to shrinking/shaping cusps?

Also, if square-form magnets are desired, the elliptical cross-section would be easier to form bends with.

C.O.

[rjaypeters]

Good thinking, been done before:

viewtopic.php?t=706&highlight=elliptical

Look for Colonel Korg's post about half-way down.

I'm aware of the potential advantage of the elliptical cross section, but I haven't done anything with it.

Keep thinking! And there are 39 pages on that thread!