Talk:Polywell

From Wikipedia, the free encyclopedia

This article may be too technical for a general audience.
Please help improve this article by providing more context and better explanations of technical details to make it more accessible, without removing technical details.
Archive
Archives
Archive 1


Contents

[edit] Lawson criterion

Has Bussard mentioned confinement time for either electrons or ions or energy? Has he mentioned the density or temperatures? How close is he to the Lawson Criteria? It would seem to me that the triple product neTτE would not improve with size because the electrons escape as they change direction. This is why mirror machines fell from favor decades ago. Paul Studier 00:11, 16 January 2007 (UTC)

A Polywell MaGrid will beat the pants off any simple mag mirror, although they do have things in common. Mag mirrors have horrid cusp losses, and the MaGrid recirculates any electrons lost to the cusps. And the magnetic confinement is electrons, not ions (which would be enormously harder). The MaGrid is the anode of a diode, with the electron emitters at about the potential of the outer walls of the machine. The electrons only want to hit the MaGrid, and the magnetic field greatly delays that. The grid behavior makes recirculation possible, the WiffleBall factor makes it trap sort of like a mag mirror but better. I can tell you that WB-6 is estimated to hold on to the electrons for on the order of 100,000 transits of the machine, and they stay at very high kinetic energy (which makes them make a potential well for the ions).
I have not seen the ion lifetimes but it is probably surprisingly high. That will depend on the density. Too high, you swamp out the ability to drive the machine and have excessive unproductive collisions, too low and the fusion rate drops (although ion lifetime is higher at low density, as this gridless electrodynamic potential well is nearly ideal for confining them). However, the fact he got it to run with such a simple puff-gas system to inject deuterium suggests the "sweet spot" is pretty large. I have little to offer that would apply directly to Lawson's criteria of temperature, density, and confinement time in a plasma in thermal equilibrium. That's not what Bussard's machine does. Something more or less equivalent goes on ... long ion trapping in a deep potential well, maintaining high peak kinetic energy as they dynamically recirculate, and high density at the point of highest, or nearly the highest, kinetic energy. That's supposed to be the easy part in this device, once the cost of hanging on to high energy electrons is dealt with.
WB-6 was no-where close to breakeven. You can get a hint of how far off by realizing the WB-6 radius was around 0.15 meters and he thinks he needs 1.5 meters to hit breakeven. While the actual output scaling is B^4R^3, his working assumption is evidently that B will rise in proportion to R, so he generally expects the fusion produced to go up as R^7, and power gain to go up as R^5. From that, I leave you to whip out the calculator (or a napkin to count the zeros, actually) and draw your own conclusions. But the point is, he doesn't think we need to sneak up on this by building 10 machines over the next 40 years, trying to marginally improve each one with theta pinches, etc. He thinks we can go straight to the proper size and it will either work or come darned close to it. And if it comes up short, a very little additional scaling up ought to work. The great thing is, this doesn't make some open-ended project that goes on for half a century. The cost of the full-sized machines is, frankly, cheap if they have any significant chance at all of working, and it should not take very long to build one at the required size. The thing would be far smaller and simpler than ITER. Frankly, I see no reason we can't support ITER and this too. Why should ITER feel threatened? Would they be afraid of it succeeding? They should be overjoyed. All those guys have the skills (after dethermalizing their fusion education) to build these things. They wouldn't be out of jobs.
We tend not to think so much of "temperature" in IEC machines, but the drive volts were 12.5 kV and the well depth about 10 kV on WB-6. That puts the kinetic energy of the deuterons at up to 10 keV (11604 K/ev x 10000 eV, or on the order of 110-120 million degrees K). But remember that this is no Maxwellianized thermal machine. Virtually all the deuts meeting in the middle are at the same kinetic energy, instead of just a few at the tail of a distribution, and there are a lot of head-on, or close to head-on, collisions possible. Also remember, it is velocity, not KE, that actually figures in rate of fusion, and head-on collisions hit like 4x the kinetic energy (i.e. twice the velocity, and KE goes as 1/2 m v^2) of either particle would do against a stationary target, as far as rate of reaction goes. BTW, that's not a violation of conservation of energy, its just that KE (moving to stationary frame) is the lookup-value for fusion cross-sections. Tom Ligon Tomligon 02:11, 16 January 2007 (UTC)
I think what most of us would like to know is some general explanation how you keep those atoms at the same kinetic energy. Even for reactions with a large cross section, shouldn't off-target collisions dehomogenize the kinetic energy levels and flush efficiency down the toilet? How do you get those atoms out of the reactor without losing too much energy? You don't have to give us nobel prize worthy answer, we'd just like to have *some* idea how this concern is addressed. 82.135.66.148 11:24, 6 February 2007 (UTC)
On each pass, as the ions approach the MaGrid, their kinetic energy and velocity drop to near zero. This makes them bunch up. In a properly-run machine, this is the only zone where the fuel ions have a condition of thermal equilibrium. In this zone, they Maxwellianize back to a low average kinetic energy. The collision crossection for this is supposedly quite high in this region, and they equalize back out on every pass. I belive Bussard mentions this in both the Google talk and the October 2006 paper, "The Advent of Clean Nuclear Fusion", page 13, second paragraph.
As soon as they leave this region, they accelerate back toward the center of the machine, a zone where density is low. As they approach the center, the odds of a head-on high energy collision go up dramatically (optimum for fusion). In the very center, collisions are from all angles, but at high energy, and high density. The ions spend about 1/1000th of their time in this high-density region, and generally don't experience enough scattering that the mechanism above can't "anneal" it out. Tom Ligon 162.84.67.130 18:33, 6 February 2007 (UTC)
Hi Tom, that explanation seems a bit lacking to me. In particular I can't get it to add up in terms of entropy. Collisions will necessarily increase the overall entropy of the ion distribution function, and this remains true no matter if they occur in the high density region or out at the perimeter. Now, if a process, or a set of processes, have as the net effect to restore the non-maxwellian velocity distribution, then it follows directly from the second law of thermodynamics that it must require an amount of work corresponding to the negative change in entropy. Now, unless you use D-D fuel ( or another fuel consisting of only one ion species ) you will necessarily have collisions which have a neglectable chance to contribute to the fusion process. For the p-B plasma you have p-p collisions as well as B-B collisions as an example. For monoenergetic ions the resulting energies energies after a collisions would range from 0 to 2E_0, where E_0 is the original energy of the ions. It thus appears to be absolutely impossible that these ions will all reach close to the same potential energy without a large input of work to compensate for the decrease in entropy that this would require. This is of course particularly true for ions which have collided in the core, as their kinetic energies before the collisions will be the highest.
Furthermore, it is obviously impossible for the average kinetic energy of the ions to decrease as they "Maxwellianize" (as you call it ) since conservation of energy would require that the potential energy is increased accordingly. For this to occur the ions would have to spontaneously convert their average kinetic energy into potential energy, and if this is to yield a monoenergetic distribution of potential energies you would most certainly see a large decrease in ion entropy without any corresponding input of work. In summary, it would appear to me that any spontaneous process which tends to counteract thermalisation of the ions, would either have to result in a large heat loss, or require the corresponding input of work. 85.230.195.223 00:25, 13 October 2007 (UTC)
Hi, Tom, and thanks for answering all our noob questions. I'm still unclear as to some of the engineering concepts for the larger device. In particular, how would the magnets be protected from the fusion products? Anyone else see any major engineering goobers with Bussard's device? Thanks! Smilindog2000 16:20, 6 May 2007 (UTC)
I'll answer this one for Tom, as I've asked him the same question myself. Short answer is that currently, they aren't. That's a definite engineering hurdle to overcome in a net-power producing machine. With the p-B11 reaction, a significant portion of the alphas WILL strike the MaGrid, causing microcrystalline damage and ablating. Also, if the coils are superconducting, the excess heat caused by the impacts could cause a quench of the superconducting state, which could wreck the reactor. Some sort of shielding will have to be devised for the interior faces of the coils, possibly with a way to reduce ablating and recapture some of the energy lost by the alphas going where you don't want them. One way to counter the ablated molecules polluting the reaction would be to use boron for the shielding. That way, any released boron would contribute to the reaction, hopefully without flooding it. This would have to be carefully done to make sure any released boron would get properly ionized, and may provide a way to limit the amount of B11 to be injected. — Blane Dabney 22:34, 9 May 2007 (UTC)
Adding a bit, if a little "original noodling" can be tolerated, here's a reality check on the scope of the problem. Bussard has, in some of his space application papers, speculated that these reactors might hit 6 gigawatts or so, at a size somewhat larger that the 4-meter diameter magrid proposed for a p-B11 demo reactor. Compare that to the "power output" of an individual Space Shuttle main engine, each of which averages, based on rate of water production, just under 6 gigawatts, for about eight minutes, using regenerative cooling by the cryo fuel to prevent damage. They have a design life adequate for many flights. The magrid surfaces in such a machine would need to be cooled on the order of what is required for a SSME (I don't have the combustion chamber diameter at hand). I would classify this as a significant engineering challenge, but not out of the realm of possibility. Terrestrial powerplants with more modest power density would be built long before any space applications, and should be less challenging. Tom Ligon Tomligon 23:44, 14 June 2007 (UTC)
I like reality checks. The biggest discrepancy between fusion exhaust and rocket exhaust is that the rocket exhaust is stone cold (eV compared to MeV). Let's try another calculation. 6 GW of fusion power in the form of alphas with an average energy of (8.7 MeV/3) over a sphere of 2 m radius gives 2.6e20 alphas per second per square meter. If we make the wall out of boron with a density of 2.34 g/cm3 and an atomic weight of 10.8 g/mol (7.7e-30 m3/atom), and assume that we have to replace the wall at the latest after 1 cm has eroded, we can sacrifice 1.3e27 boron atoms per square meter. Now for a big assumption. How many boron atoms will be ablated by each alpha? Given that the energy of the alpha is more than a million times the energy of evaporation, probably a heck of a lot. If each alpha ablated just one boron, we would have to replace the wall after 5.1e6 s, i.e. every two months, which would be a pain, but might be doable. For a more realistic ablation ratio, you would be forced to provide some sort of continuous replacement of the exposed surfaces. And you can't afford to let a significant fraction of the ablated fuel enter your plasma, either (although you won't be able to stop it). The replacement rate of boron fuel is one per three alpha particles, so if your grid is 99% transparent, you still can't afford to produce more than 30 ablated atoms per alpha. How's that for reality? --Art Carlson 08:32, 15 June 2007 (UTC)
One better, for 100 MW fusion power, I calculate boron use at about 1.6 milligrams per second, and production needs to be regulated closely (the performance envelope is narrow regarding fuel density and it is necessary to keep the virtual anode height fairly low). Using boron as an ablative coolant would very likely choke the reactor if they come off faster than that. While I'm generally familiar with sputtering and related phenomena, I frankly have no idea what shield materials might be expected to do when hit by 3 MeV alphas. A large sputtering load of anything is probably bad, especially high Z materials. Much better if the alphas simply dig into the material and make heat, which would need to be removed by a cooling system. (I assume these shields will require periodic replacement.) If this is, in fact, a killer problem, tokamaks are similarly affected, as material sputtered from their walls would poison the reaction. Tom Ligon Tomligon 18:58, 19 June 2007 (UTC)
I'm afraid sputtering is too complicated, so I can't do much more than say it seems to me likely that the sputter yield will be much greater than unity. Sputtering is indeed a problem that has received a lot of attention in tokamaks, but they differ from polywells in many respects. For one thing, the alphas in tokamaks are contained by the magnetic field long enough that they cool down to the temperature of the plasma, which is only a few eV next to material surfaces. For another, the erosion is to a large extent countered by redeposition, which would be nearly absent in a polywell. --Art Carlson 19:44, 19 June 2007 (UTC)

[edit] Future Work

Some of the information in the 'future work' section is really exciting, but can we please have some citations? The bit about Bussard's widow and the team in Santa Fe is one long piece of original research.JulesVerne 22:08, 8th November 2007 (GMT)

Like I said earlier, the following is great news for fusioneers, and I'm hugely pleased at the apparent success of the work:

Dr. Bussard passed away on October 6, 2007. His work will be continued by the staff of physicists he was able to assemble at EMC2. Dolly Gray, who co-founded EMC2 with Bussard in 1985, and served as its president and CEO, has helped assemble the small team of scientists in Santa Fe. The group includes Rick Nebel, Jaeyoung Park, both physicists are on leave from LANL; Mike Wray, the physicist who ran the key 2005 tests, and Kevin Wray, who is the computer guru for the operation. The latest device, WB-7, is currently under construction at a machine shop in San Diego and will be shipped to Santa Fe, where a small group of scientists is setting up a testing facility. The device, like previous ones, was designed by engineer Mike Skillercorn.

Mike Skillicorn. I fixed it. Tomligon (talk) 02:37, 3 February 2008 (UTC)

Suggestions have been made to have a multi agency review of the the results and schematics to encourage timely public release of all findings and documentation.

BUT unfortunately not one citation to back it up! How do we know that WB-7 is being constructed in San Diego? Which reputable source states this is the case, and how can it be verified? Is it original research from people involved in the project? Where is it from? .JulesVerne 12:15, 2nd January 2008 (GMT)

I didn't add that content, but I know the source is an article in the New Mexican, a New Mexico newspaper. I'll try to dig it up and cite. —Preceding unsigned comment added by 205.234.189.1 (talk) 20:30, 23 January 2008 (UTC) Never mnind, it's already there. It's the last external cite. 205.234.189.1 (talk) 20:38, 23 January 2008 (UTC)

[edit] image?

A diagram of the field and the confined plasma would be helpful —Preceding unsigned comment added by 129.2.40.144 (talk) 22:16, 26 November 2007 (UTC)

[edit] Three neutrons, visible light

From the paper The neutron counts were in the range of 3 counts per test, for the closest counters, which calibrates out to about 5E5 fusions during the device operating time. So I changed it back. Paul Studier (talk) 20:18, 9 December 2007 (UTC)

It is not detection of three neutrons total, it is three per test. 24.13.35.175 (talk) 23:57, 21 January 2008 (UTC)

If you think visible light production is not related to fusion energy and should not be cited as supporting evidence to neutron counts as an indication fusion may have occurred, you may get a big surprise if you go outside and look up. ;) 205.234.189.1 (talk) 19:09, 23 January 2008 (UTC)

And you can go outside and look at a bonfire. --Art Carlson (talk) 19:45, 23 January 2008 (UTC)
I didn't see a pile of wood in any of the WB-6 pictures, but I'll check again.205.234.189.1 (talk) 19:53, 23 January 2008 (UTC)
I am sitting under florescent lights that have glowing plasma but no fusion. Paul Studier (talk) 20:32, 23 January 2008 (UTC)
Well, set up some shielded neutron counters on your desk, and if you get counts that work out to 5E5 fusions over a 1/4 millisecond just as visible light is produced, please let us know, because the two together probably mean something.205.234.189.1 (talk) 20:44, 23 January 2008 (UTC)

Here is the source doc on the PMT events:

The measured data from these tests shows DD fusion neutron production (Fig. 7) of about 5E4 neutrons over a period of about 0.2 msec (less than the data rate interval), which also shows the emitter current of injected electrons (Fig. 8) to run at about 4-40 A during this short pulse period of fusion generation. .This peak pulse period is also indicated by light output measurements from the photomultiplier tube detectors (Fig. 9). The PMT showed a rise to peak output as the internal machine neutral gas was fully ionized, a flat-top during the onset of the external glow discharge, and a rapid falloff as this condition was passed. The actual rise was certainly faster than the data rate showed, so that at the peak, the edge electron density was a maximum, the full well depth was established, and DD fusion was taking place. Beyond this time, the potential on the machine dropped (Fig. 10) as external arcing (from the tank walls and feedthroughs) took over, the external current rose to very high values (Fig. 11), and the system discharged and shut down.

So presumably they understand the difference between light from glowing plasma and light from something that is not glowing plasma.

Fourth, the PMT registered a large pulse for 4 of the 5 tests, indicating a large light output, in the center of the machine, at exactly the time of the counts. There is nothing there in the PMT’s field of view to cause an arc (which usually is the culprit for noise).

It proves little or nothing by itself, but it seems worth mentioning along with the neutron counts. —Preceding unsigned comment added by 205.234.189.1 (talk) 21:05, 23 January 2008 (UTC)

Actually, it's not worth mentioning. Nobody ever suggested that the light was not from a glowing plasma. Even an arc would be a glowing plasma. (I think he is worried about showing that it is not an arc because then you might be getting the neutrons from beam-target fusion in the walls.) Not a word in the passage you cite suggests that the light emission is driven by fusion. In fact, it is clear from other numbers in the article that it is not. If the energy gain scales as the 5th power of the radius, and you would need a factor of ten bigger radius to make a power plant, then the gain of the device tested must be about 10^-4. That would mean that fusion adds 0.01% to the energy you already put into the machine. The light he saw if definitely not coming from fusion, so why mention it? --Art Carlson (talk) 22:07, 23 January 2008 (UTC)
Nobody? Well, somebody mentioned bonfires as a source of light ;) Anyways, OK, they aren't arguing that the fusion produced the light, just that the fact they happened at the same time is an indication that something other than noise was happening at the moment the neutrons were recorded. As the source says, that's relevant because the other most likely explanation for those neutron counts is noise, esp. given the small number of tests. So I'm not clear on what your objection is, though, if we all agree the fusion did not produce the light. Are you arguing the source is wrong in that claim that the visible-spectrum light pulse being concomitant with the neutron counts is more evidence fusion occurred?
From the archive, we see that the drive voltage is 12.5 Kev at a current of 14 amps giving 175,000 watts. The neutron rate is 10^9 neutrons/second giving a power of about 0.001 watts. Not enough to light up much of anything. The drive power exceeds the fusion power by a factor of 175 million times. Paul Studier (talk) 22:49, 23 January 2008 (UTC)
That's interesting, you and Art come up with energy gain many orders of magnitude apart. Using Art's .01%, I would get 17.5 watts, which might conceivably be visible. But Art's probably just being too optimistic. —Preceding unsigned comment added by 24.13.35.175 (talk) 00:09, 24 January 2008 (UTC)
OK I get 1E+9 fusions x 17.6 MeV per fusion x 1.6E-19 eV per joule = .00282 joules/second = .00282 watts, so we're in the same ballpark there. Does seem pretty dim. 24.13.35.175 (talk) 00:29, 24 January 2008 (UTC)
That is interesting. I would say these calculations call Bussard's claim into question, even accepting the R^5 scaling of gain, that a full scale device would only need to be 8-10 times bigger than his experiments. Not that it makes much difference, if you can really get that scaling. I was too lazy to calculate the power, like you did, but that is the more direct and accurate approach. Remember also that this flash last less than a millisecond, making it still harder to see. --Art Carlson (talk) 08:20, 24 January 2008 (UTC)
That was something I wondered about too. I'm not sure how he gets to that based on those numbers. I'll ask around, as that is an interesting discrepancy and people interested in Polywell are mostly just sitting around waiting for Nebel's results anyway. Also, I'm told a sensitive PMT may produce 1E6 electrons per photon (remember, the pulse was visible *spectrum* light detected by the PMT, not something someone actually saw) so it's not impossible they detected a very dim light from secondary effects of fusion. But reading through the notes again, it seems more likely their point is just that the light pulse was indicative of the conditions that were likely for fusion, making it additional evidence that the neutron counts were not just noise as they happened simultaneously with the light pulses.
I get 100MW gain at radius x 38, based on the WB-6 results (38^5 x 1e9 neutrons / 1.76MeV per joule x .5 fusions per neutron = 111,563W). I don't remember hearing that the plant was going to be ~40x larger than WB-6. So maybe he was assuming other efficiencies, or doing the calculations based on expected p-B11 rates/energies.205.234.189.1 (talk) 16:35, 24 January 2008 (UTC)
Oops, no, I think maybe we have the wrong number there. The 1e9 would be the fusion power, not the gain, wouldn't it? The gain would be the power minus the loss, but obviously no attempt was made to determine the loss. So using the r^7 scaling for power, I get 14MW power @ rx10, and we get to 100MW at about rx14. I'm not sure what the gain is, because of course we have no loss numbers for WB-6, but we could infer from r^7 - r^5.205.234.189.1 (talk) 17:31, 24 January 2008 (UTC)

[edit] NSD claims

Regarding this edit: "Claims" is OK, since I wouldn't take their hype at face value either without independent verification. I thought "offers for sale" would cover it, since they are offering it, whether they can actually deliver or not. It's just that there are now "my word against his" statements in the article, with a formulation slightly in favor of NSD because of the word "however". I just wanted to offer an explanation that they are maybe not referring to the same conditions, or even the same device. Unfortunately we don't have enough information to really decide what is going on. That's the difficulty of writing articles about topics that are not adequately covered by secondary sources. --Art Carlson (talk) 11:04, 24 January 2008 (UTC)

My mentioning NSD was an attempt to put Polywells fusion rate in some context. If anyone can find a reputable source for the neutron rate of a fusor, then we should use that instead. Paul Studier (talk) 21:26, 24 January 2008 (UTC)
I found a reference from U of Wisconsin. Fewer neutrons but much more reliable, IMHO. Results comparable to Polywell without the complications of the magnets. Paul Studier (talk) 01:43, 25 January 2008 (UTC)

[edit] Polywell geometry

I don't understand one bit of the article. It says that polyhedrons should have an even number of faces at each vertex, but don't the cube (used for all iterations of the polywell to WB-7) and the dodecahedrom (future WB-8) both have three faces at each vertex? —Preceding unsigned comment added by 131.111.200.200 (talk) 02:17, 1 March 2008 (UTC)

Both the cube and the dodecahedron have to be rectified (fully truncated). This is mentioned in the article, but a bit indirectly. On 14 Feb an anonymous user added the precise names but was reverted 2 days later. Would it have helped you to understand the concpet if you had read cuboctahedron instead of truncated cube, or icosidodecahedron instead of truncated dodecahedron? I suspect not, seeing as you apparently skimmed over the "truncated" part anyway. Any other suggestions? --Art Carlson (talk) 07:06, 1 March 2008 (UTC)
Perhaps they could be given a brief description of what they are and then linked to cuboctahedron/icosi... Such as "a torus on each face of a cube, resulting in an approximation of a cuboctahedron..." Kevin Baastalk 16:04, 1 March 2008 (UTC)
Or inserting "fully" would do, much more concisesly. :) Kevin Baastalk 16:06, 1 March 2008 (UTC)

[edit] Wrong summary about thermalization?

Bussard contested that thermalisation can occur, and claimed that the probability of collision is lowered by the high speed through the core of the plasma, while at the rim, it has lower energy and hence does not thermalise to any major degree. By the time that thermalisation would have occurred, the electron would have been lost from the system.

a) It does thermalize at the rim and b) it's not about electrons thermalizing.

As far as I understand it this would be more accurate: According to Bussard the high speed and therefore low density of the ions in the core makes thermalizing collisions very unlikely, while the low speed at the rim means that thermalization has almost no impact on ion velocity in the core. 82.135.13.167 (talk) 01:01, 14 April 2008 (UTC)

I agree that our summary of Bussard's arguments is hard to understand and probably wrong. Then again, I would say the same about his arguments as presented in the reference we cite. If you think you can make sense out of Bussard's paper and present it in a way useful for a lay reader, please do. Otherwise, it might be better to just say that Bussard dissents and refer the reader to his paper for the arguments. --Art Carlson (talk) 07:47, 14 April 2008 (UTC)
Regarding these two possibilities, I believe the thermalization (i.e. collision) probability would be something like density^2 * cross section, so both factors would be relevant, though density would scale faster. Regardless, the sentence begins "according to Bussard...", so we should be careful that whatever is put there is "according to Bussard"; if he used one or the other choice of words, we should use that. Kevin Baastalk 21:42, 27 April 2008 (UTC)
The Coulomb cross section drops with the fourth power of the velocity, so it usually wins. Remember, too, that the polywell intends to increase the density in the center through convergence of the ion "beams". Anyway, I agree that in the end we need to follow Bussard's statements (if we can figure out what he's trying to say). --Art Carlson (talk) 06:32, 28 April 2008 (UTC)
From the paper cited in that section:

Ions spend less than 1/1000 of their lifetime in the dense, high energy but low cross-section core region, and the ratio of Coulomb energy exchange cross-section to fusion crosssection is much less than this, thus thermalization (Maxwellianization) can not occur during a single pass of ions through the core. While some up- and down- scattering does occur in such a single pass, this is so small that edge region collisionality (where the ions are dense and “cold“) anneals this out at each pass through the system, thus avoiding buildup of energy spreading in the ion population (Ref. 14). Both populations operate in non-LTE modes throughout their lifetime in the system. This is an inherent feature of these centrally-convergent, ion-focussing, driven, dynamic systems, and one not found (or even possible) in conventional magnetic confinement fusion devices.

Looks to me like cross-section is the winner. Kevin Baastalk 17:13, 28 April 2008 (UTC)
It's difficult to decipher exactly what he thinks the two species are doing, independently and in relation to each other. I thought that the core, being a focal point, would be dense in electrons, hence "virtual cathode", but from "At the edge the electrons are all at high energy where the Coulomb cross-sections are small, while at the center they are at high cross-section but occupy only a small volume for a short fractional time of their transit life in the system." it seems that it is not electron dense. Or is he saying that, although the ratio of electrons in the core to those outside the core is small, the ratio of the volume of the core to the volume of the outside is much smaller? That's the only way i can see the core to have high electron density while at the same time have electrons not spend very much of their time in it. Kevin Baastalk 17:31, 28 April 2008 (UTC)
Or is it that since electrons have much lower mass than ions, but the same charge, their moment of inertia is much lower and thus they will change direction a lot quicker in response to a voltage gradient, while ions will tend to plow through the conflicting gradient because it takes them longer to turn away from it -- allowing for a greater electrostatic buildup (and thus density). Thus making the core much more ion-dense than electron-dense? Kevin Baastalk 17:43, 28 April 2008 (UTC)
But ions would also be traveling much slower than electrons - so they're turning radius would be about the same and they'd actually have about the same inertia. - it seems like the magnetic field is the main thing here because that does produce a different turning radius in ions vs. electrons. The electrons are going to follow the magnetic field lines while the ions will not (will do so much less) - so that's what will pull the electrons away from the center faster than the ions. The ions just follow along the stream of electrons for the ride - accelerating to match their (net) velocity to neutralize any voltage gradients - until the magnetic field lines cause the electrons to turn sharply - sling-shooting the ions into the core.
Since now the core is much more positive it will attract some electrons to stray off from the magnetic field lines and "hang out" in the core where their movement will be dominated by electro-static forces; by trying to fill the ever-moving electro-positive holes in the voltage gradient. Meanwhile ions are circulating in the faces and out the corners of the core (or vice-versa). Am I getting this right? Kevin Baastalk 18:03, 28 April 2008 (UTC)

[edit] Magnetic fields

Art, not sure why you think magnetic fields are concave on the outside, or why it would matter. The electrons just follow the field lines; it doesn't need to be concave for them to follow the field lines around the outside and re-enter. TallDave7 (talk) 19:13, 16 May 2008 (UTC)

Helpful illustration.

http://www.fusor.net/board/view.php?bn=fusor_theory&key=1174706460&pattern=recirculate —Preceding unsigned comment added by TallDave7 (talk • contribs) 19:24, 16 May 2008 (UTC)

Where to start. Thanks for the illustration, but I already understand the geometry of the polywell. I didn't put in the bit about the concave/convex fields. It's been in the article for some time. That's not really the right terminology. I believe it's a question of whether the gradient of the plasma pressure and the gradient of the (scalar) magnetic pressure point in the same direction or in opposite directions. Next, charged particles do not "just follow the field lines". There are plenty of drifts and other effects that determine the paths of particles, so you shouldn't be too sure you understand what's happening at first glance. You reinstated this clause: recirculating back into the machine along field lines due to their attraction to the positively charged magnetic grid (magrid). This is saying more than just that the electrons follow field lines. It would suggest that the electron density is highest near the grids and lower both farther inside and farther outside. I don't know that anyone is claiming that, but he would also have to take into account that circulating electrons actually spend less time in regions of high positive electric potential, and the effects of magnetic mirroring. Furthermore, the densities and loss rates will be a balance of some sort between leakage out of the cusps, recirculation back into the cusps, losses in the exterior region, and losses to the grids and grid supports.This is a hell of a lot of OR. What can we say that is either undisputed or can be attributed to secondary - or at the very least independent - sources? I don't think the current statement about recirculation falls in that category, so I will revert it (by and by). But maybe you can make an alternate suggestion? --Art Carlson (talk) 21:53, 16 May 2008 (UTC)
P.S. I went to some effort to describe the differences in the magnetic configuration between tokamaks and polywells in this edit. I admit it may be a bit too technical. Did you have any other specific objections? Why did you revert it instead of improving it? --Art Carlson (talk) 21:58, 16 May 2008 (UTC)
"I believe it's a question of whether the gradient of the plasma pressure and the gradient of the (scalar) magnetic pressure point in the same direction or in opposite directions." -- Okay, sure, that's fine for the confined electrons, but I don't see where that's an issue in regards to recirculating electrons that have escaped through the cusps. They're just trying to get to the Magrid, so for them the field is convex. "Next, charged particles do not "just follow the field lines". There are plenty of drifts and other effects that determine the paths of particles, so you shouldn't be too sure you understand what's happening at first glance." -- Of course, they would follow a spiral around the line, bang into each other, etc. It's a rough description, the rough description Bussard and Ligon have used, and as you noted earlier we should try to rely on Bussard's description. Bussard specifically designed the machine with the electron gyroradii in mind when he spaced the grid coils because of the need for recirculation. "It would suggest that the electron density is highest near the grids and lower both farther inside and farther outside." -- Yes, some who follow the subject believe that to be the case. "the densities and loss rates will be a balance of some sort between leakage out of the cusps, recirculation back into the cusps, losses in the exterior region, and losses to the grids and grid supports" -- Yes, that's the general idea. "This is a hell of a lot of OR." -- Not sure what you're referring to here. "I admit it may be a bit too technical. Did you have any other specific objections?" I have no objections to how technical it is, I reverted it because you removed the mention of the positively charged Magrid, which seems central to the difference between Polywell and a tokamak. TallDave7 (talk) 00:57, 17 May 2008 (UTC)
"P.S. I went to some effort to describe the differences in the magnetic configuration between tokamaks and polywells Why did you revert it instead of improving it?" OK, that is a good point; that addition is actually very good. "But maybe you can make an alternate suggestion?" -- I unreverted your addition (pasted your text back in) and just added the Magrid stuff to the end.
I'm still not exactly happy, but let's work on it.
  • I think the convexity of the fields is not that important. The original text suggested that it was a black-white comparison. I have made a case that both types of fields play a role in both machines, and anyway transport and micro-instabilities are much too complex to be reduced to a single characteristic. Do you think it would be OK to eliminate the discussion of convex fields entirely?
  • The other things about flux surfaces, cusps, and variations in the field strength are important and undisputed (whatever conclusions you want to draw from them). These points practically force a comparison between the polywell and mirror machines. Wouldn't it be helpful to make the similarities and differences more explicit? The configuration that comes closest is the minimum-B mirror, also called yin-yang or baseball coils. Unfortunately there doesn't seem to be any information on this in Wikipedia. In contrast to the polywell, there is a magnetic axis and no field null.
  • I am still bothered by the comments on recirculating electrons and their attraction to the magrid, for several reasons. To begin with, there is no such thing as a positive grid in an absolute sense. It it only positive in relation to something else, in this case the outer vessel. The plasma potential is determined by complex processes, so it is not easy to say what the plasma potential is relative to the grid. Another (related) problem is that the electron density would normally adjust itself to shield the grid, so that the electric field would be significant only within a few Debye lengths of the grid. There, it would be perpendicular to the magnetic field and the primary result would be an E-cross-B drift parallel to the grid surface. (How big is the Debye length under nominal polywell conditions?)
  • One more comment related to the electron losses. Either there is significant electron pressure in the region exterior to the grid or there isn't. If there is, then you have to worry about lots of things like plasma pressure driving instabilities in the region of concave fields. (Think "collective processes", not individual particle orbits.) If there isn't, then you would expect the rate at which electrons re-enter the interior to be much smaller than the rate at which they leave.
  • My comment about original research meant to say this: We are discussing a lot of physics here, which is not really our job. If we have to discuss the physics of a particular edit because we can't find any secondary sources that do it for us, then there is a good chance that it would be better not to include the topic in the article at all.
--Art Carlson (talk) 09:52, 17 May 2008 (UTC)
I'm not trying to come up with anything original, I'm just trying to cite what Bussard has said about how the machine operates. TallDave7 (talk) 17:37, 17 May 2008 (UTC)
I hope you guys don't mind me interjecting for a third opinion:
  • The convexity of the magnetic field is important on the inside of the magrid for two reasons 1) it provides magnetohydrodynamic stability, and 2) it provides the "wiffle ball" effect. Where either of these effects absent, the machine would not work. If something that makes or breaks the system isn't important, I don't know what is.
I agree 24.13.34.40 (talk) 17:28, 17 May 2008 (UTC)
  • over my head.
  • sure, by positive, one means positive voltage gradient towards. This is such a common abbreviation in electronics that it's vernacular. But as I understand it, the electrons recirculate because of the magnetic field lines, not because of the charge - and this charge is created by the electrons being trapped by the magnetic fields. If we were talking about ions, then i'd go with charge. but electrons, i'd definitely go with magnetic fields. And as far as i recall that's what the literature says.
Both, according to Ligon (see picture). IIRC Bussard said much the same, though I'd have to dig around for a cite. TallDave7 (talk) 17:37, 17 May 2008 (UTC)
  • I thought you said concavity/convexity of the magnetic fields wasn't important. I don't know what you really mean by this talk of "electron pressure". This is a high-vacuum plasma environment. Any force applied by electrons is the result of force supplied by either 1)magnetic fields, 2)electric fields, or 3)thermalizing collisions. So where one would consider "electron pressure" one considers magnetic pressure, electric pressure, and thermal diffusion - plus inertia. This doesn't seem very complex to me. on the outside of the magrid you have electrons and ions. electrons have tiny gyroradii compared w/ions, and the magnetic fields are spread apart so they have low interactivity. so the electrons are going to spiral around the field lines and back into the chamber, assuming 1)they don't collide with something (such as a metal surface), 2) their angle of incidence isn't beyond critical (in which case they shoot out), 3)they aren't overpowered by an electric field. (ultimately an electric field would effect the critical angle.) electrons are going to be more dense inside the magrid so the force from the electric field would push the electrons outward. (neglecting here the ions, which would just neutralize this field, lessening the effect). So if you're talking about "electron pressure" as "force supplied by electrons due to spatial differences in density", it's outward, not inward. while magnetic pressure is - well - in circles, actually. My point of all this is that electron recirculation is a product of the lorentz force, not the coloumb force. Kevin Baastalk 15:32, 17 May 2008 (UTC)

Hold that thought. I'm probably going to be incommunicado until next Wednesday or Thursday, but I'll take up the discussion again then.--Art Carlson (talk) 16:08, 17 May 2008 (UTC)


OK guys, here's a cite:

"Thus, in order for a Polywell to be driven in the mode described for the basic concept, open, recirculating MaGrid (MG) machines are *essential*. This, in turn, requires that the entire machine be mounted within an external container surrounding the entire machine, and that the machine be operated at a high positive potential/voltage (to attract electrons) relative to the surrounding walls. Note that this was the electric potential configuration used in the earliest MG machines, the WB-2 device, that proved internal magnetic trapping of electrons, called the Wiffle-Ball (WB) effect. And in the first proof of Polywell fusion reactions, in MPG-1,2, and in fusion production in the later devices, WB-4, 6."

http://www.askmar.com/ConferenceNotes/2006-9%20IAC%20Paper.pdf

He may or may not have been right, but this is how the machine is supposed to operate.

TallDave7 (talk) 17:37, 17 May 2008 (UTC)

I have boldly rewritten and renamed this section. If I have gone too far, feel free to revert, but please copy my text to this talk page so we can work through it. If you see some merit in the new version, try to improve it instead of reverting it. --Art Carlson (talk) 13:25, 22 May 2008 (UTC)