non-uniformly charged magrid?

[KitemanSA]

from my simulations, there are two regions of electron containment, one that covers most of the region inside the magrid, but stays just shy of it, and contains fairly thermalized electrons. it's looks like a "cloud", and does not have much visible variation in density from outside to inside.

the other region is a very small hollow spiky sphere of very cold electrons in the very center of the magrid. this area is MUCH denser than the thermalized electron cloud. and presumably these electrons are held in place by magnetic mirroring from multiple angles (from the multiple coils).....

Does your mag field react to the electron pressure? Does it "blow up" like a balloon as you feed in electrons? Of perhaps the better analog would be flatten out like a bunch of balloons

Your descriptions suggest that the wiffleball is a tiny thing in the center, but I was under the impression that it is actually almost at the MaGrid wall by the time the unit is operating. The wiffleball is supposed to turn the HOT (~10keV) electrons, not contain the cold ones.

And the act of inflating it is supposed to squeeze off the point cusps until they are very tight, reducing the rate of escape, no?

[happyjack27]

Does your mag field react to the electron pressure? Does it "blow up" like a balloon as you feed in electrons? Of perhaps the better analog would be flatten out like a bunch of balloons

Your descriptions suggest that the wiffleball is a tiny thing in the center, but I was under the impression that it is actually almost at the MaGrid wall by the time the unit is operating. The wiffleball is supposed to turn the HOT (~10keV) electrons, not contain the cold ones.

And the act of inflating it is supposed to squeeze off the point cusps until they are very tight, reducing the rate of escape, no?

instead of trying to describe it, why don't i just show you:

http://www.youtube.com/user/happyjack27 ... GFmtQJrLq8

notice i have the electrons start in the center instead of shooting them in there with an electron guns. that's because i haven't calibrated the velocity of my virtual eguns yet.

you'll see in that video me slowly increasing the electron density, while keeping the mag field strength at around 7 tesla on a 3m-radius magrid. the charge on the magrid is nearly off, at about 10^-14 coloumbs per meter. it starts to really break containment at around -10^-7 coloumbs space charge.

i think to get the tightening cusps i'd have to increase the charge on the magrid, to geometrically counteract the mag fields, making the cusps less pointy so that more electrons deflect more inward.

[hanelyp]

Charge on the magrid shouldn't have much effect on particles inside it. As far as electrons inflating the magnetic field like a balloon, it should be enough to account for the magnetic field generated by moving electrons and add that to the total magnetic field.

[93143]

It sounds like you're operating in the wrong regime.

You can't just start the electrons at zero and expect a wiffleball to form. They'll just stay in the centre, constrained by the magnetic field.

To form a wiffleball, you need a lot of fast charged particles to push the magnetic field back. Only after the plasma charge density has reached equilibrium will the electrons be slow near the centre.

Also remember that the net charge is only about 1e-6 of the total charge count in the plasma (I believe that was for a WB-6-class machine). You need a lot of fast electrons and ions (they should be fast in different regions, but proper injection should handle that), with a very small excess of electrons; the whole plasma contributes to the diamagnetic effect that forms the wiffleball.

The electrons guns are just thermionic emitters; they drop the electrons into the positive potential well created by the magrid charge, which is where the electrons get their energy and why the whole thing works. And you shouldn't need all that much accuracy to get the electrons inside; they should just spiral down into the cusps (using a physically realistic density, or at least collision rate, may be helpful in getting them off the field lines). If you try to circumvent this by starting electrons at almost zero in the centre, you're missing the whole physics of how they got that way, which is actually very important, and your simulation won't recover that for you.

I should also note that the magrid doesn't have a constant charge - it has a constant potential. The charge on it increases to maintain the potential difference between the magrid and the surrounding components (e.g. the electron emitters) as the negative charge of the wiffleball builds up.


Very interesting simulation, though. I wish I had the time to do some simulating of my own (I have a couple of prototypes in Matlab sitting around), but I'm quite busy with my Ph.D. research right now...

[happyjack27]

It sounds like you're operating in the wrong regime.

higher net charge per coil current leaks more electrons, while a lower ratio has smaller net charge and thus shallower potential wells. and i would presume "right" means highest power out per power in. though i'm just judging visually right now. we'll see how it looks once i add the phase space view.

You can't just start the electrons at zero and expect a wiffleball to form.

duh. you need a mag field for that.

They'll just stay in the centre, constrained by the magnetic field.

and that's a problem?

To form a wiffleball, you need a lot of fast charged particles to push the magnetic field back. Only after the plasma charge density has reached equilibrium will the electrons be slow near the centre.

incorrect.

Also remember that the net charge is only about 1e-6 of the total charge count in the plasma (I believe that was for a WB-6-class machine). You need a lot of fast electrons and ions (they should be fast in different regions, but proper injection should handle that), with a very small excess of electrons; the whole plasma contributes to the diamagnetic effect that forms the wiffleball.

while some electrons will inevitably be fast, preferablly they are cold, esp. in the center. the opposite holds true for the ions.

The electrons guns are just thermionic emitters; they drop the electrons into the positive potential well created by the magrid charge,

the potential well is negative

which is where the electrons get their energy

you mean they use the charge on the magrid to accelerate. yes that's one way to get the electrons into the center - to make them climb the negative potential well.

and why the whole thing works.

no, that is not what makes it work.

And you shouldn't need all that much accuracy to get the electrons inside; they should just spiral down into the cusps (using a physically realistic density, or at least collision rate, may be helpful in getting them off the field lines). If you try to circumvent this by starting electrons at almost zero in the centre, you're missing the whole physics of how they got that way, which is actually very important, and your simulation won't recover that for you.

i have simulated the brute force approach, and while in time it does start to form a wiffleball -- or what you would argue is not a wiffleball -- it is very slow and VERY inefficient. while it's true that you don't "need" to aim your eguns for the center, you will get a much deeper potential well and fewer electrons thermalized and lost per electron you fire in if you do.

I should also note that the magrid doesn't have a constant charge - it has a constant potential. The charge on it increases to maintain the potential difference between the magrid and the surrounding components (e.g. the electron emitters) as the negative charge of the wiffleball builds up.

as far as my simulation goes, to calculate the eelectric field around each wire segment, i need to start with coloumbs per meter, just like to do the same for each particle i need to know it's charge in coloumbs.

if i'll need to vary this throughout the sim, well i suppose i shall get a clearer picture once i get my phase space view up and running. but unitl then, well, it's not really a big deal.
[/quote]

Very interesting simulation, though. I wish I had the time to do some simulating of my own (I have a couple of prototypes in Matlab sitting around), but I'm quite busy with my Ph.D. research right now...

thanks. i plan on making it even more interesting

[D Tibbets]

Actually, you do need high energy to form the Wiffleball, or at least the Beta=1 condition that is synominous with the Wiffleball.

The formula is something like:
Beta= B/ n*eV [BAD Formula,BAD- please ignor]
B= magnetic strength, n= density, eV=electron energy in eV. Also, there may be a nonrhythmic function in there somewhere. I'll have to look it up.
The actual formula is closer to:

Beta=8(pi)nE/B^2

n is density of electrons, E is energy in eV

It doesn't make any difference what the charge/ potential is on the Magrid once the charged particles are inside. Where it comes into play is in accelerating the electrons on there way to the magrid cusps (from the outside) These electrons are new from the E-guns, or recirculating. If you introduce the electrons in a simulation through some other mechanism, you have to give them this initial kinetic energy (on the border). This derives the potential energy in the center. It seems that just placing a dense population of electrons in the center at low energy, and accounting for the Coulomb repulsion would end up with the same result, but I am uncertain.

[EDIT - In addition to the energy of the electrons (potential or kinetic, you need to provide for the radial predominance of the kinetic energy. IE: you need to set the direction / vector of the initial electron population as radial ( not just a collection of a dense random direction bag of electrons in the center ). Once operating you could allow Maxwellian forces to apply, but you need to carefully consider the location of the electrons within the system, as this effects the thermalization interactions- this is described in one of the links in the next post. ]

Bussard described (briefly) two ways to create a Wiffleball. I'll have to look it up too.

Dan Tibbets

[D Tibbets]

Some papers you should carefully read, if you haven't already done so.

Forming and Maintaining a Potential Well
in a Quasispherical Magnetic Trap

The Advent of Clean Nuclear Fusion:
Superperformance Space Power and Propulsion


Electron Current, Beta Limit Line Operation
and Power Balance in WB Mode

Phenomenological Modelling of Polywell/SCIF
Multi-Cusp Inertial-Electrostatic Confinement Systems

Preliminary Study of Inertial!
Electrostatic!Fusion "IEF#
for Electric Power Plants

I think all of these can be found at:

http://www.askmar.com/Fusion.html

Dan Tibbets

[happyjack27]

thanks for all this info dan.

hopefully i can get my phase space evolution view up soon so i can get a much clearer picture. then i'll start soaking up that material and see if with that knowledge and the more informative views in my simulation i can get the simulation running in a more ideal operating regime.

[happyjack27]

Actually, you do need high energy to form the Wiffleball, or at least the Beta=1 condition that is synominous with the Wiffleball.

...

Beta=8(pi)nE/B^2

n is density of electrons, E is energy in eV

is that electric potential energy, kinetic energy (radial, presumably), or both? simply by starting off close together they start with high pe.

In addition to the energy of the electrons (potential or kinetic, you need to provide for the radial predominance of the kinetic energy. IE: you need to set the direction / vector of the initial electron population as radial ( not just a collection of a dense random direction bag of electrons in the center )

i'm starting them off at 0 ke. their mutual repulsion accelerates them radially from there. given that there's about 10k of them, their initial position distribution is pretty uniform so there's not much discrepancy to provide for electrostatic axial acceleration. in the beginning, at least. when they start getting into the cusps it's a different story.

[93143]

i'm starting them off at 0 ke. their mutual repulsion accelerates them radially from there.

Okay, uh... I must admit I somehow glossed over that in my thoughts about your simulation. It was late at night; what can I say...?

My comments stand, however, in general - if you aren't simulating what the real system does, your simulation will often find a way to deviate from how the real system behaves.

1) The magrid needs to be positively charged; ie: generate a positive potential well wrt the electron emitters (NOT guns; they are thermionic emitters, and thus don't need aiming) that exceeds the depth of the negative potential well inside the device. This is what provides electron confinement. Without it, fast electrons will escape the system easily, and only the slow ones will be left behind in the centre...

2) Are you just using electrons? You'll need a ridiculously high magrid potential to get the necessary plasma pressure that way...

...I'll let you read up on it.

[happyjack27]

2) Are you just using electrons? You'll need a ridiculously high magrid potential to get the necessary plasma pressure that way...

it varies. before i could do all electrons, all ions, or mixed. now i have fined-grained control over the populations while the simulation is running.

re: electron guns. i understand if you just use guns that don't give any initial velocity to the electrons (thermonic?), you have to use the magrid potential to accelerate them. in that case the combination of their distance from the magrid and the magrid potential determine their ultimate speed when they pass through the center of the coil.

alternatively if you had guns that gave the electrons an initial velocity, well that would add linearly to their speed and thus you could place them closer and/or reduce the magrid potential.

in any case you don't want them to fast either, else they'll shoot right past and out the other side. ideally their energy would be exactly equal to the electric potential energy at the center, plus maybe a little extra to compensate for being pushed off the axis a little on the way there.

i've noticed since the ions dont like the positively charged coils, the ones far from the center will tend to gravitate towards the center of them, providing a nice track of positive charge to keep incoming electrons from scattering too much.

on the other hand, by the same logic, shooting electrons in from the corners might provide better focus, since the coils are closer together there.

[93143]

No, you've got the wrong idea. The magrid potential has to be entirely responsible for the electron energy; otherwise the escape rate is too high.

This is an electrostatic confinement device. The magnetic field is only a booster.

Also, the electrons should probably be fast in the centre until the well nears equilibrium. This is how the real system works. If you try to compensate and have the first ones be slow at the centre, you aren't pushing the magnetic field back hard enough and you'll have trouble forming a proper wiffleball. Don't worry about them leaving; the magrid charge will pull them back in.

How fine-grained is your population control? IIRC, a WB-6-class device is supposed to have a ratio of electrons to ions of about 1.000001...

[happyjack27]

How fine-grained is your population control? IIRC, a WB-6-class device is supposed to have a ratio of electrons to ions of about 1.000001...

well there are two logarithmic sliders for particle population. both range from 10E-14 to 10E0. at that range, the increment is about E0.06

right now one of them is net charge, and the other is total charge. (sum of absolute value of charge). so to get a ratio of say 1.001 i set the total charge 3 units higher than the net charge.

i suppose i could change that to ion/electron ratio. such that net electron charge / net ion charge = 1+10Ex where x ranges from -14 to 0, with an increment of 0.06. the first slider would then be population difference, and the second would be population ratio, then the raw population counts can be calculated from there.

[93143]

I don't see a reason to change it if it works. (Also, if you did ever want to do a sim with only electrons or only ions, it would be bad if the ratio went singular...)

Can your method handle large local deviations from the specified ion/electron ratio? I'm thinking yes, but technically there are ways to do a particle-tracking simulation that would be unable to deal with this...

[happyjack27]

I don't see a reason to change it if it works. (Also, if you did ever want to do a sim with only electrons or only ions, it would be bad if the ratio went singular...)

Can your method handle large local deviations from the specified ion/electron ratio? I'm thinking yes, but technically there are ways to do a particle-tracking simulation that would be unable to deal with this...

ya sure. it's an all pairs n-nody particle-in-cell simulation. there are 14k particles, 6 parts electrons, 1 part protons, 1 part boron11. they simulate a larger population by making each particle appear to the others as many particles with the same position and velocity. so even though it's a limited resolution approximation, the hydro-electro-magneto-dynamics (or whatever. the fields and their changes) will be that of a larger population, though not as finely meshed / uniform.