Talk-Polywell.org

The biggest problem with a Farnsworth-Hirsch fusor--which the Polywell is designed to solve--is the fusor's physical grid. The fuel ions in an F-H fusor tend to impact the grid before they have a chance to fuse, leaving them with insufficient energy for fusion. Furthermore, the grid is relatively fragile; it's unlikely a wire mesh grid could be designed to survive, say, the power output levels needed for a practical power plant.

But it seems to me that the research done on the polywell effort, whether or not it ever succeeds in producing a practical power-generating polywell, may enable a practical F-H fusor.

Specifically, Dr. Nebel's team has calculated that if the polywell magnets are strong enough (10T?), the magnetic fields will even deflect the fusion-product alphas away from them; the alphas will bounce around inside the polywell until they finally find a cusp and escape, and the MaGrid will be spared the collision-induced spallation and heat load problems.

So imagine a Farnsworth-Hirsch fusor--except that instead of using a wire mesh for its grid, it uses an arrangement of 10T superconducting magnets pretty much identical to a MaGrid, except it would be negatively charged (since it's still essentially an F-H fusor) instead of positively charged.

Like regular F-H fusors, the positively-charged fuel ions would fall toward the center of the fusor grid and (assuming no ion-ion collisions) out the other side. But unlike regular F-H fusors, the strong magnetic field would eliminate the ions' collisions with the grid--negating the first big problem with F-H fusors. And just as with a 10T polywell, the fusion-product alphas of this fusor would also escape without hitting the grid, greatly reducing its heating, thus negating the second big problem with F-H fusors.

But unlike a polywell, there would be no mixing of positive ions and free electrons, since the electrons are confined to the (electrically conductive) grid--which would keep Bremsstrahlung energy losses in check.

Has this idea been floated before? Does it have any obvious flaws? I make no claim to being a physicist, and I don't know how to evaluate or critique this idea. I actually expect there's a fatal physics flaw in here somewhere that I'm not getting, but I need a little help in figuring out what it is.

It's hard to tell now, at least until we will know if the Polywell concept does work and how does it work.
Bremsstrahlung energy losses might or might not be an issue in Polywell, and more surprises might come out when a detailed report of actual experiments will be finally issued for pubblic scrutiny. That is IF it will ever be issued.

Wellcome aboard.

from a stastiscal point of view, for productive power level densities, and the proximty of the central magnetics to the fuel/fusion region, they would have to be prety darn strong magnets I would think to accomplish the task and not get cooked. It also does not account for the supports/power/cooling hookups.

A magnetically shielded central cathode has been kicked around. I don't know if any one has ever built one with sufficient magnetic strength. Even modest strength magnets would seeming,y improve results to a degree. A cathode will emit electrons with sufficient voltage, especially if heated up (hopefully minimized with magnetic shielding). So there would be some electrons, I have no idea of the amount.

The main problem with this approach (if it worked) is scaling. The density of the plasma within the Polywell magrid is several thousand (or more) times greater than the background pressure. And, you cannot increase the background pressure to compensate due to arcing concerns, and other issues.
The Wiffleball trapping factor increases the density from perhaps 0.0000001 atm to 0.0001 atm. This 1000 fold increased density results in a million fold increase in the fusion rate. To get a magnetically shielded fusor to reach the same output, it would have to have 1 million times the volume to compensate. Confluence issues may modify this somewhat on both sides.

A 1 GW Polywell might be 4 meters in diameter. A Tokamak might be 40 meters in diameter, and a magnetically shielded fusor might be 400 meters.

Losses are important, but so is power density.

Dan Tibbets

in a way, a polywell IS a fusor with a magnetically shielded cathode. the magrid is the cathode. it has a bias voltage with reference to the plasma and the outside walls, and it's electromagnetically shielded. the only way to have a fully magnetically shielded cathode is to use electromagnetics (cause static magnets have field lines that run right into the magnet) and have the cathode conform to said fields. that means the only way is to have a collection of closed curves -- i.e. a "magrid".

so then the only parameters left to play with are the shape and orientation of the curves, the strength of voltage bias relative to magnetic field strength (current), and your other standard variables (size of grid (cathode) vs. container, size of container, rate of ion injection, etc.)

"polywell" and what one might call "a fusor with a magnetically shielded cathode" (for having said parameters closer to a typical non-magnetically shielded fusor) are just two possibilities of these parameters. the most distinguishing differences being the relative size of the grid, a restriction on the orientation of the electromagnetics, and the ratio of bias voltage to current (e-field to m-field).

so essentially, a polywell IS a fusor with a magnetically shielded cathode, just with some of the parameters optimized for more power density and fuel efficiency. (the WB effect being an optimization over a very nonlinear region of parameter space)

timothy.d.power wrote:The biggest problem with a Farnsworth-Hirsch fusor--which the Polywell is designed to solve--is the fusor's physical grid.

Such statements always makes me sad. It's like no-one is really listening, or thinking about it for themselves, just repeating some tall tale someone else has said.

There is no 'big problem' with the fusor. It does what it does well enough, which is to keep ions running along beam tracks colliding with the background medium, with which they may fuse.

Fusion is a random process. Most of the time the ions scatter off those collision with the background medium and loose energy by heating up the background. This is the process of 'thermalisation'. The grid makes next to no difference to this. The losses to the grid are small compared with the other losses in a fusor. Think about it... a standard 6" fusor requires around 400 watts to be pumped in to make around 1million neutrons/s. Now, just consider a 400W bulb. How much light gets radiated with an input of 400W? And yet a well-designed fusor's grid doesn't even glow red (some do, if the grids are a bit wonky, this its true, but never 400W worth).

So.... where's all that input energy going??.... It ain't going into the grid, 'cos the grid ain't emitting 400W of power!!

There have been successful fusion neutron-emitting gridless fusors and, y'know what? They work pretty much the same as gridded fusors. Negligible difference in behaviour.

The standard two-grid fusor doesn't do net energy gains because it can't - just like ANY such beam device that doesn't prevent thermalisation of its ions. Gridless fusors have been built and operated and.... same thermalisation problems. Attempting to remove the grid from a standard fusor is the biggest red-herring in the history of electric fusion.

Prevent thermalisation and... well, then you'll be on to something....

i agree that loss to thermalization is big. but i think there's a calculus nuance one must consider in charged particle loss to grid.

the rate (dx) of loss to grid is going to be proportional to the density of particles around the grid (x). something like dx = Cx (that's a _partial_ derivative, of course) e.g. when you stop pumping in, it's going to decay exponentially (/logarithmically).

now the simple what goes in must come out principle (law of conservation) tells you that, thermal losses aside, the net loss on the primary loss component (in this case the grid) is going to be pretty close to the rate of input (ion/electron injection). i.e. it will reach an equilibirum. so don't be suprised that the loss to the grid is always only a little less than what you're pumping in.

now look at ion lifetime in the fusor. maybe 1,000,000 orbits before it runs into the grid, on average. then your total ion volume's going to be about 1,000,000 times what you're pumping in.

now remove that primary loss channel, you know, the one that only lost about 1/1,000,000 of the ions per orbit. if that extends the lifetime of an ion/electron in the fusor 1,000-fold, you've effectively increased the density per unit electron/ion pumped in 1,000-fold. e.g. your volume is now 1,000,000,000 units. i.e. you now have 900,000,000 more eV in your fusor. when only 1/1,000,000 (0.000 000 001) eV per orbit were being lost to the grid before.

but that 900,000,000 eV doesn't mean a grid would be piping hot. sure. if you dropped one in at that point it would melt right away. but if you could keep it together long enough to absorb the energy until it reached equilibrium, you'll see that now again you're back at 1M volume and the temperature/eV loss of the cathode has dropped back down 100,000%.

Well, all I can say is that both plasma electrodes, needle electrodes and single loop electrodes have all been tried by amateurs, and they are all much of a muchness. If the grid made that much difference, then wouldn't you think you'd see differences in such experiments? In fact, such 'solutions' tend to work, if anything, a little worse than a 'nice regular grid' because self-organising beams don't tend to form up. The beams in a fusor strike the outer shell causing electron emissions, generating ions [at the top of the potential] and, thus, strongly maintain ion production.

A quick calc on energy loss wrt MFP suggests the ions have at best a few hundreds to thousand cycles in them (as any such fast ions would in a unit-Pa vacuum, that didn't receive fresh 'input' energy).

If anyone is actually interested in this stuff, best to read Appendix E.2.2 if Todd Rider's thesis. That should give you a few clues on how to make an IEC device work.

chrismb wrote:Well, all I can say is that both plasma electrodes, needle electrodes and single loop electrodes have all been tried by amateurs, and they are all much of a muchness. If the grid made that much difference, then wouldn't you think you'd see differences in such experiments?...

no. i wouldn't. you still have the same primary loss channel. it just has a somewhat smaller surface area. so it's a multiplier on your dx, i.e. it's a different "C" in dx = Cx. in the end when it reaches equilibrium the difference in current from the electrode vs. what you're pumping in isn't going to be very significant. and likewise the ion lifetime probably isn't going to be much different. certainly nothing on the order of a thousand fold.

and like you mention later it doesn't focus the plasma as well so you're going to lose a lot more w/a worse density then you could possibly seek to gain by slightly less loss surface area.

you have to eliminate all ion/electron loss channels completely. any small loss channel is going to suck the life(times) out of the ions/electrons and drop your equilibrium density, voltage, and temperate down orders of magnitude.

So are you trying to say that if you plotted 'loss' on y-axis against 'grid intercept area' on x-axis then you think you would just see a horizontal line?

chrismb wrote:So are you trying to say that if you plotted 'loss' on y-axis against 'grid intercept area' on x-axis then you think you would just see a horizontal line?

i'm saying you actually have to plot at least 6 axis. loss, grid area, present ion/electron volume (or density or what have you), ion/electron lifetime, and input (ion / electron injection). and what you'll get is a system of differential equations. (some of them rather fixed).

when those equations reach equilibrium, you'll have essentially an energy equation where the two sides are equal. and on one side you're have mostly ion/electron injection and current drain from electrictiy source, and on the other you'll have mostly heat lost to container and grid (and since it's densest and hottest in the center of the chamber, you're certainly not going to lose a lot of heat energy to the container, relatively speaking.) and current drain to ground / electricity source through the grid, but mostly the latter. and, all things considered, the loss area will have a fairly negligible effect on what that equilibrium will be.

i suppose my main point is there will come a point where doubling your energy input isn't going to double your ion density because the grid will just bleed off that input energy faster so the resulting density isn't going to change proportionally.

and though that point may be slightly higher with a thinner grid, anything above it---and you will no doubt have to go _way_ above it---you're going to have to put in exponentially more energy to get the same increase in density, and you'll never break even that way.

you have to find a way to stop that process completely (or almost completely) if you want to break even. hence the virtual cathode, recirculation, and it being absolutely critical to minimize if not eliminate ion / electron loss channels.

Does it have any obvious flaws?

If you're shielding the cathode grid, you're not making it transparent, you're just putting a big bumper around it. That will tend to defocus the ions.

But even if you had a fusor with a perfectly transparent cathode grid, you'd still have issues with space charge limits and Coulomb scattering that would prevent you getting near net power.

http://en.wikipedia.org/wiki/Fusor#Use_ ... wer_source

The Polywell is basically an ETW with a shielded anode grid and magnetic confinement of the virtual cathode.

http://en.wikipedia.org/wiki/Polywell#P ... son_fusors

The big question mark at this point is whether the electron confinement in a PW machine is good enough. Cusp confinement probably isn't good enough, but so-called "wiffleball" confinement might be. It's hard to model an answer because you have things like 3D shear flows that are computationally intensive. WB-8 will give us some idea how the loss scaling with B to .8T looks.

As mentioned, in the Polywell, the magrid is actually a shielded anode as in the Elmore Tuck Watson (ETW) variation of the fusor. The cathode is 'virtual' and is driven by the injection of energetic electrons towards the center. . I mentioned the plasma density that is claimed for the Polywell. Another factor, even if you protect the wire grid from ion impacts, in a typical fusor the electrons stream straight to the vacuum vessel walls, so electron losses are still a problem, even if ion energy losses are reduced. The ETW overcomes this somewhat by having a virtual central cathode formed by introducing electrons outside of an anode grid so that they are accelerated towards the center. Magnetically shielding this anode gets you part way there. The geometry of the magnetic fields to form tight cusps (Wiffleball) gets you closer to the goal, and finally, the recirculation of the otherwise unavoidable electron leakage to the walls finishes the race. At least that is the claim. The near spherical geometry is also important for confluence issues and (I think) thermalization/ annealing issues and bremsstrulung issues. There are a lot of road blocks involved. It is remarkable that the Polywell seems to navigate around them.

Concerning glowing wire cathodes in fusors. The glow( heat) of the cathode is not due to current flow- perhaps 20-50 mA of electrons from the grid to the walls. A thin wire can handle those currents with impunity. It is the accelerated ions (and charge exchanged or upscattered neutrals) striking the cathode that heat it up and eventually sputters material off of it till it is eaten up (or melts).

Dan Tibbets

Many thanks for all the responses to my layman's question! I think I'm now seeing some of the problems with a net-power producing magnetically shielded gridded fusor:

--It doesn't solve the thermalization problems; after not too much time in operation, the fusor would be choked with ions too low in energy to fuse, while higher-energy ions would be upscattered eventually to escape from the system, taking their energy with them.

--The ion density is highly dependent on the level of charge on the cathode, but there's an upper limit on that charge beyond which arcing occurs. As a result the fusor can't achieve high enough ion densities; as a result a fusor-based power plant with decent output would have to be unreasonably large.

--The magnetic shielding might protect the grid, but it would serve to deflect the falling ions away from the fusor's center, "defocusing" them--resulting in fewer energetic collisions with other ions (and thus fewer fusions).

Have I about got it?