simulation run for octahedral fusor

[hellblazer]

Attached are some gifs of some ion tracks and the potential contours for a typical amatuer fusor. It's ideal, so your real world potentials will have spherical abberations considerally different than the simulation.

All grid units in millimeters. Simulation is a 300 mm spherical chamber, octahedral grid radius is 100 mm. (a 12 inch chamber with an 8 inch grid).

The simulation is a spherical octachedron inner grid held at potential of -30,000 volts and a spherical ground electrode representing the vacuum chamber.

The octaheron is rotated 45 degrees around the X, Y, and Z axis, so that the X axis goes straight through the center of two faces of the octahedral grid, and the center of the octachedral grid.

The ions enter in from position 5,5,-140 (from the upper left in the attached gifs). They have an initial energy of 1 eV along the X axis. The ions are spread with +1,1,1 added to the previous ion's position.

The first attached gif is the potential energy well of the XZ cross section, with the five ion tracks.

The second gif is the potential contours for the XZ cross section.

The third gif is the ion track of a single ion which originates at 0.05mm, 0.05mm, -140 mm. It has an initial energy of 1 eV parallel to the X axis. Translated, this is a single ion that is on a parallel track to the X axis, 50 nano meters away. This would represent ions in an ideal, tightly focused beam, perhaps.


As you can see, the saddle of the potential gradient is a bulge, rather than a dip. The result of this bulge is that anything but an ion going precisely down a radial line will end up orbiting closer and closer to one of the grid electrodes and eventually ending up as a power loss. I can't count the precise number of orbits the ion takes, but it's certainly less than 50 - say order 10^2.

Things get worse if you take into account beam interaction between the ions. Not able to simulate the positive space charge that builds up in the center from the ions. (i.e. multiple potential well structure).

[DaveC]

Nice plots Hal -

Sometime back, in a different thread.. I described the potential well you show there. I used Poisson (2D) to model the cage... and the potential well at the cage center was a few % below the cage voltage, as I recall.

While these plots don't deal with the non ideal vacuum, field emission and thermionic emission, you can write up a program for adjusting the initial energies to reflect electrode temperature and field changes, when ions hit the electrodes. Not trivial, but, if I understand the capabilities, definitely possible. then makes the plotting almost a minor thing, compared to the complexity of handling the source description. Scattering from non-ideal vacuum, is, in principle, also possible, but would require a statistical evaluation at each ion step.. to determine if a gas molecule was hit and if so, which direction the ion went next.

Actually, when you look at the paths, it is a wonder that anything happens!!

I need to sit down and do this for the reflex cylidrical configuration too.

Dave Cooper

[hellblazer]

> Actually, when you look at the paths,
> it is a wonder that anything happens!!

Yep.

I've been doing some simulations on Bussard's design, and the presence of the "virtual" grid composed of trapped electrons considerably helps. The trapped electrons are much more spherical, resulting in a far, far better focus. It's not a point focus, but the number of orbits an ion can make numbers in the millions. And the initial positions and momentum an ion can take and still not hit an electrode after millions of orbits is amazingly huge.

Still have to model my magnetic circuit - SimIon isn't great for that - so we'll see. I may be able to get by with no potential on the grid, only using trapped injected electrons to form the virtual grid.

[DaveC]

Hal -
How are you modeling "trapped" electrons? Are you setting up an electron gun to aim at the center, and then also aiming an ion beam at the same region? That seemed to be what Farnsworth originally described. Also, my earlier simulation (was only 2D ) indicated the positive central region went away as the number of cage electrodes increased. Thus if a uniform screen shell were used, there would be no virtual anode, from the potential, only from the electrons passing through and concentrating near the center as they scattered away.

I don't yet know how to set it up, but I believe you can arrange to record "hits" at any specified coordinate , which could indicate what the collision frequency is in different regions of the fusor.

Dave Cooper

[guest]

I was wondering, what application do you use for your simulation and where did you get it?

[grrr6]

>Actually, when you look at the paths, it is a wonder that anything happens!!

Not just the paths, but in this simulated case, the ions that go straight down the middle only get what looks to be about half the grid potential to fall through, as most of the rest is concentrated right around the grid wires.

[hellblazer]

SimIon

My understanding is that they have great student pricing for this.

[hellblazer]

Doh! Didn't even notice this. Makes sense, though, don't it? Inverse square and all that...

Now I see the violence inherent in the system...

I mean, now it makes sense that measured energy of ions in the Australian's experiments showed 20% of the grid potential...

Yi.

[hellblazer]

Right now, I just use an ideal spherical grid in SimIon. This is assuming that I can trap the electrons with Bussard's magnetic cage. I'm modeling the magnetic circuit with Radia.

See Bussard's first patent for more detail...

The shorter version is that you build a magnetic circuit which only has point cusps. Then you inject electrons into this circuit. The electrons form a virtual cathode. All you have to do is make sure the plasma charge balance is negative to keep it operating.