magrid configuration brainstorming

Discuss how polywell fusion works; share theoretical questions and answers.

Moderators: tonybarry, MSimon

Mike Holmes
Posts: 308
Joined: Thu Jun 05, 2008 1:15 pm

Post by Mike Holmes »

Hi, total layman here. But I'm reading with a lot of interest. The geometry question is fascinating.

At first, seeing the triangular-sided figure on page one that somebody posted, I was tempted to suggest an isocahedron, as it would have a very spherical shape (hence why they're often used to model globes, etc). But then I realized that this would probably actually increase the size of the vertex gaps, proportional to the size of the rings. And I'm not sure if it would make sense in terms of how you'd charge each ring.

But then it occured to me that the problem is that the well isn't "capped" at these corner locations. What if you put another grid outside of the inner grid, but with a geometry that worked around that of the inner grid. Such that it was open where the inner grid pulled in, and pulled in where the inner grid was open?

I can think of two geometries that might work with this. The first is that I note that a cube has 6 sides, and 8 vertices. And the octahedron is precisely the opposite, with 6 vertices, and 8 sides. You could arrange an octahedron outside of the cube such that the sides of the octahedron that caused inbound pressure were located above the cube's vertices, and the octahedron's vertices were above the cube's sides.

Another geometry that occurs to me is to simply use two tetrahedrons made of rings. Each has four sides, and four vertices. The outer is aranged so it's rings are over the inner tetrahedron's vertices.


I'm sure it's quite likely that these ideas won't work, for technical reasons of which I'm unaware. Perhaps these ideas have even been considered and discarded already - apollogies if they have been. But since this is a brainstorming thread, I thought that I'd toss the ideas out. Even if it turns out that these can't work, they might get somebody's head stimulated toward coming up with a superior geometry outside of the box.

If you'll pardon the double entendre.

Mike Holmes

P.S. If I recall correctly, Farnsworth didn't have a degree in physics or engineering - or any degree at all - when he invented his fusor. :-)

blaisepascal
Posts: 191
Joined: Thu Jun 05, 2008 3:57 am
Location: Ithaca, NY
Contact:

Post by blaisepascal »

Mike:

The thing to remember is that without the use of magnetic monopoles you have to have the same number of magnetic field lines going out as coming in, no matter what geometry you use. Putting another shell around the polywell won't eliminate the cusps, but it'll push them around and make them more complicated to compute.

You could eliminate the cusps with magnetic monopoles, but unfortunately they are prohibitively expensive to make in any useful quantity.

Mike Holmes
Posts: 308
Joined: Thu Jun 05, 2008 1:15 pm

Post by Mike Holmes »

Sure, I wouldn't expect that such a model would eliminate loss entirely. Merely that the deformation of the fields might be such that loss might be reduced. I understand that it might be difficult to model such a geometry, however. It just seemed intuitive to try to provide inbound momentum from all angles.

But as with many things in physics, sometimes intuition isn't a good predictor.

Mike

MSimon
Posts: 14334
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Post by MSimon »

but unfortunately they are prohibitively expensive to make in any useful quantity.
Isn't that the truth.
Engineering is the art of making what you want from what you can get at a profit.

cuddihy
Posts: 155
Joined: Sat Jun 30, 2007 5:11 pm

Post by cuddihy »

Ok, Tombo I think I see a potential problem with your coil formation, most visible to me on

Image

It's not a problem "inside" the device so much as realizing that a recirculating machine means that both "inside" and outside flows of the machine matter. We'll have to wait to see plots for anything definitive (Go Dr. Mike!), but one of the key features of a recirculating polywell like Wb-6 or 7 that was missing in WB-5 is that field lines don't touch metal.

At first glance, that's true of your coils too. But draw a single equipotential field line "surface" around the coils. On the two-circuit one, you'll see that in many ways you've essentially recreated the "equator" problem of the original mirror machine, only this time at the lead entry point instead of at the equator.

The geometry's different. But it should have close to the same effect: a natural, rapid flow of electrons into the vessel entry point for the leads similar to at the equator of a two-coil mirror.

WB-6 and 7 avoid this problem by the multiple turns of each coil. The leads still go in a similar geometry. But the large field lines of the coil that bounds the wiffleball will shield the far smaller field lines following the leads straight into the vessel wall.

I could be completely wrong here, perhaps a single turn geometry can be jiggered that sufficiently shrinks that surface to still allow a wiffleball geometry. Or maybe my concerns are a red herring.

Just a consideration.
Tom.Cuddihy

~~~~~~~~~~~~~~~~~~~~~
Faith is the foundation of reason.

Solo
Posts: 261
Joined: Fri Jun 29, 2007 12:12 pm
Location: Wisconsin

Post by Solo »

@Mike Holmes: Welcome to the forum! I think that Dr. Bussard mentioned using a configuration like you suggested: another set of coils to plug the cusps. I'm not sure what the verdict is on the usefulness of that method, but one thing is that more coils means more alpha-particles intercepted by the coils, which we want to avoid.
Mike Holmes wrote:But then I realized that this would probably actually increase the size of the vertex gaps, proportional to the size of the rings.
Exactly. For a given size machine, more coils means more cusps (one cusp at each vertex and each coil center) and each cusp has a minimum area, so more coils = more cusps = more leakage.

drmike
Posts: 825
Joined: Sat Jul 14, 2007 11:54 pm
Contact:

Post by drmike »

Tom - I think a basic "right hand rule" proves you are right. That's why I know my code has bugs :) Just put your thumb along the wire and the fingers show the field direction. At the entry point, we can think of the hoops as connected since all the currents entering equal those exiting. It won't be the same as a circular coil - all the triangular faces will be mirrors. The field at the entry should be almost zero so it'll be a strange looking cusp (relative to circular coils). We'll see when I get the plots to work....

jmc
Posts: 427
Joined: Fri Aug 31, 2007 9:16 am
Location: Ireland

Post by jmc »

MSimon wrote:
but unfortunately they are prohibitively expensive to make in any useful quantity.
Isn't that the truth.
If by prohibitively expensive you mean physically impossible then yes.

jmc
Posts: 427
Joined: Fri Aug 31, 2007 9:16 am
Location: Ireland

Post by jmc »

@Tombo

The holes aren't closed by the wiffle ball. Even in the presence of the wiffle ball the holes will still be one gyro radius in diameter for the point cusps and wide for the line cusps.

In mirror mode you have additional problems in that collission and loss cone instabilities will eject out the electrons after about 1 electron-electron collission time. This is because they already are in the magnetic field so they are mirror confined which, in a collissional plasma at least can be worse that cusp confinement.

Wiffle balls don't "seal" cusps, the loss area of the cusps are as good as it get in terms of confinement and cannnot be eliminated. But maybe reduced by increasing the magnetic field, decreasing the larmor radius.

My greatest concern about the polywell, is whether or not the line cusps really can be eliminated in a practical device. If they can't I suspect it will be very touch and go as to whether a compact version of the device Busard proposed will work at all. Its true that in a fully recirculating device electrons emerging from the cusps will just go back in again, but ions are not completely confined either as there will be earthed equipotential corridors out of the cusps towards the vacuum vessel. This will represent a very real energy loss and while it may be manageable in the presence of only point cusps, in the presence of line cusps it could be prohibitive.

MSimon
Posts: 14334
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Post by MSimon »

jmc wrote:My greatest concern about the polywell, is whether or not the line cusps really can be eliminated in a practical device. If they can't I suspect it will be very touch and go as to whether a compact version of the device Busard proposed will work at all. Its true that in a fully recirculating device electrons emerging from the cusps will just go back in again, but ions are not completely confined either as there will be earthed equipotential corridors out of the cusps towards the vacuum vessel. This will represent a very real energy loss and while it may be manageable in the presence of only point cusps, in the presence of line cusps it could be prohibitive.
This is only a problem if the ions get enough energy through collisions (up scattering) to escape. If there is a natural mechanism that prevents or greatly reduces that then there is no problem. Ions are electrostatically confined.

Two methods have been proposed. Thermal annealing and beam bunching. There is no way to tell from data publicly available which is the right answer. Or even if both are true in varying proportions. Or neither.
Engineering is the art of making what you want from what you can get at a profit.

TallDave
Posts: 3140
Joined: Wed Jul 25, 2007 7:12 pm
Contact:

Post by TallDave »

ions are not completely confined either as there will be earthed equipotential corridors out of the cusps towards the vacuum vessel.
Hmm, how so? Isn't the bottom of the well always at the center for ions?

drmike
Posts: 825
Joined: Sat Jul 14, 2007 11:54 pm
Contact:

Post by drmike »

In general, yes. But the electron density shields the MaGrid from the ions, so there may be potential zones near ground (or close enough). It will be very interesting to find out!

TallDave
Posts: 3140
Joined: Wed Jul 25, 2007 7:12 pm
Contact:

Post by TallDave »

Oh yeah, the Debye length issue.

I guess I always assumed you either had the Magrid pushing them in or the electron well pulling them in, no matter where they were. I suppose we don't have enough understanding of how the plasma behaves to say for sure.

Maybe Nebel and co. will have the measurements to tell us something about the dynamics there.

tombo
Posts: 334
Joined: Fri May 09, 2008 1:10 am
Location: Washington USA

Post by tombo »

Cuddihy,
draw a single equipotential field line "surface" around the coils .
You could be right. But I’m not quite sure what you are pointing at.
Which equipotential surface are you looking at?
I see the fields of each adjacent pair of leads adding in the space between them (except on the one line in the very center.)
Also the leads are very close so the fields between them are very intense.
I think the fields will look somewhat like the ones at the center of Indrek’s picture of 2 opposing coils ("electron recirculation" discussion Fri May 16, 2008. 8:48 pm)
the large field lines of the coil that bounds the wiffleball will shield the far smaller field lines following the leads straight into the vessel wall.
I thought it would be ok because the leads have the same field as the coils so they shield themselves with the same current that the coils do.
A multi-turn machine would only self shield the parts of the leads close to the magrid.
The field lines around the leads run perpendicular to the leads not to ground.

But, yes one of my concerns is just what happens in between the leads.
My hope was that the aspect ratio of the escape path is so long and narrow that they can’t go far before being pulled back in by the E-field.
I can’t tell how much effect the E-field will have within the 4-lead volume. It will be partially shielded.
The models should tell us a lot.
Also I’m not sure what the long thin path means in velocity space.
My theory on it is that the corners of the octahedron will close by the whiffleball effect sooner than the centers of the faces where the field is smaller.

A work-around could be to cross the leads somehow where they enter/leave the tetrahedron.
But that reduces the possibility of using the lead axis as electron/ion gun paths.
It also messes up the symmetry, but that might be what we need to do to tie a knot in the fields.
It also won’t work at the other 4 corners of the octahedron where the same problem exists (if we are talking about the same issue here.)

Another work around is to cross all the corners like a cyclone fence.
But this makes the fields very messy by adding a 3rd dimension to keep track of at each corner.
(I had this in my back pocket for another occasion but will lay it out now.)
Building the whole magrid like a cyclone fence could be used to create an arbitrarily large number of coils with the even number of faces at each corner.
This idea is totally half baked but might show some promise.
There are several obvious hurdles for it to jump, but that is for a different discussion.

Jmc,
By closing the holes of course I mean reducing them as much as possible.
I used the word “closed” to try to cut down the length of my already too wordy posts.
They get way too long if I try to qualify every statement.
Did I say “seal” if so I was being sloppy. Sorry.
Closed is a relative thing. This is the real world after all.

Further
My understanding of the loss mode in the original cusp machines (2coil) is that
there is a plane of zero field at the equator where the gyro radius goes to infinity.
I don’t see where we have zero fields in these devices (except within the conductive plasma.)
All points are surrounded by either a coil or a virtual coil.
Between tangent coils (also bends at triangle corners) the fields add and are concentrated so even higher.
Even if there were a zero field zone the gyro radius could be no larger than the spacing between opposite coils on a cube on or an octahedron. (this could be a reason to go to lots of small coils.)
So even then they can’t go very far before being pulled back in by the positively charged Magrid.
I don’t think we will have any true cusps. Not any more than one isolated coil has a cusp along its axis.

Does an isolated coil have a point cusp along its axis?
Then do 2 parallel wires have a line cusp along the plane halfway between them?
If so, then we have line cusps between the tangent coils.
But, that is where both the models and the experiments show the least leakage.
I think we are Shiny.
-Tom Boydston-
"If we knew what we were doing, it wouldn’t be called research, would it?" ~Albert Einstein

drmike
Posts: 825
Joined: Sat Jul 14, 2007 11:54 pm
Contact:

Post by drmike »

The easiest way for me to think about coils and fields is to draw circles around the wires. Just like an electric circuit needs a complete loop to work, and magnetic field needs a complete loop to exist.

If you make a "cyclone fence" with wires, the magnetic fields will cancel in the small hole between the connecting wires. That's way too big a leak.

Cusps are not "zero", they are places where "circles" from other points on a wire (or different wires) meet. So they are regions of high field pointing in a specific direction. The center of a coil is cusp because the circles from opposite ends of the coil meet. The line between coils is a cusp as is the corner. In all those places, the field points in one direction, and it is definitely not zero.

Post Reply