Playing with Permanent Magnet Magrids

[D Tibbets]

The picture below shows some permanent magnet variations. Using my surplus Microwave oven magnetron magnets I have built the first example. (poles facing inward. In this configuration a solid round magnet may be better as the inward facing pole results in a small point/face cusp, instead of a ring cusp with the ring magnet. If the poles can be placed on the radial sides of a ring or solid magnet the cusps would not reach the magnet surfaces till after the tightest region of the cusp was passed. With positively charged plates on the inward facing surface of the magnet (magnetically shielded) there may be some recirculation though I doubt it would come anywhere close to that obtainable with spaced electromagnets. On the plus side these magnets are up to ~ 1.3 Tesla in strength. This is about 13 times WB6 strengths, which should increase density/ fusion rate ~ 169/28,000times higher higher with everything else being equal. Of course, everything else is not same, but this shows that there is probably significant potential for confinement (Wiffleball) studies, etc. Permanent magnets in this configuration couldn't come close breakeven, but they may allow orders of magnitude improvements over typical gridded fusors, possibly making a high neutron output machine suitable for various uses. Operation should be much simpler and cheaper than electromagnets for the same results. Of, course, if used for longer than brief tests, the magnets would have to be cooled to maintain temperatures below the Curie point. (~ 350 degrees C).
In two dimensions the solid magnets look good, but the third dimension is the E-W or equatorial plane and the fields would probably be compromised Ring magnets would not suffer from this problem.
The illustrations were created with a time limited trial version of Vizimag.


I did not see an example of a complete ring magnet with radial poles (one facing in all the way around the ring while the other points out), but presumably one could be built with sub assemblies.An example of a magnet that could be assembled into a complete ring to form a ring magnet with poles on the sides (radial surfaces).

http://www.magnet4less.com/product_info ... ducts_id=1










Image of ring magnets with face poles in my 'Pressure Cooker' Demo fusor is below. This is somewhat similar to WB1 except a real central wire cathode was used.




Turning the face poled ring magnets sideways has been suggested by others. I have done this in the past, see image. Perhaps having the horizontal sides with 6-8-12 ring magnets turned sideways with the top and bottom ring magnets with the faces towards the center would confine better than all faces pointing inward. Images ot this shows the shielding of the edges with prominent cusps into the sides. I don't recall if the magnets were arranged with N poles all facing in the same direction or if they were opposed. I the poles are facing the same way and the spacing is close enough there may be good shielding all the way around, but only in the horizontal mid plane. Perhaps stacking one layer on another would accomplish something(and look like the Michelin Man), but I doubt it would be better than with the simple face pole inward truncated cube like WB6 It looks like there is some magnetic insulation in the mid plane, but making a near spherical confined space would be improbable. At best, it might be hour glass shaped.




*This could have been placed in the Magrid Brainstorming thread, but it is getting too long.

Dan Tibbets

[chrismb]

I think you know I have warm regards for this experiment of yours. I think the whole approach of sticking a grid in such a configuration has the potential to 'kick start' investigations as to what such a thing might do if it really did create a well depth in the middle.

I would say there are a number of rather wide-assumptions you're making there, and I'm guessing you know it.

Incidentally, in a 2.4 GHz magnetron, the magnets are there to generate a ~950 Gauss field - right in the middle of them. So the suggestion of 1.3 tesla is extremely far out. Those are ferrites, and even neodymiums that do have a surface field of 1.3T will have fields that die off to next-to-unmeasurable in a very short distance. You'd be looking at a few 100 Gauss (say, 0.05T max) most anywhere away from those magnets by more than a few mm.

[D Tibbets]

Yes, it is mostly speculative.
The magnets are [EDIT] 950 Gauss? That is two to three times as strong as my unfounded assumptions. As far as 1.3 Tesla strengths, I was indeed thinking of the niobium magnets.
I in no way expect such permanent magnet arrangements to compete with electromagnets in either potential strength or confinement efficiency at any strength. But, starting at ~ 10X the strength of WB6 (a better comparison may be WB4 with ~ 1300 Gauss fields). This excess strength would compensate some for other short comings, possibly to the extent that significant fusion gains may be possible over naked wire cathode only gridded fusors. If there can be some density advantage, possibly some improvement in charged particle density within the grid compared to the neutrals, fusion gains might be significant. If a modest (?) ten fold increase in density can be achieved, this might increase the fusion rate by as much as 100X. This is way short of the claimed Polywell Wiffleball trapping factor (it would be closer to the intermediate cusp confinement mentioned as being ~ 50-60 fold). Also, of course electron losses would be proportionatly greater than the Polywell, but also, the electron confinement may be better than the naked fusor (little to no confinement.
Judging from early Polywell research, it is indeed difficult to acheive deep potential wells without excellent electron confinement. I wonder if having a physical cathode could mitigate this- it works in a naked fusor.

As for achieving high Bets, the formula is ~ Beta= something * (B field strength / density * temp) . So a weaker magnet strength is compensated by lower temperature and internal density.

I'm uncertain how much the neutral contribution would be. Without buying a diffusion pump I am limited to perhaps 40-50 Micron pressures at best.
If the neutrals are not too evil, useful testing of internal density gains might detectable photometrically (how Bussard, etel did it), Visualization of the glow discharge may mimic Wiffleball effects. Also, if there is some moderate gain over naked fusors, detectable fusion may occur at lower voltages. It is easy to get 3-4 thousands volts in a ~ 50 micron vacuum level. In a nakad fusor this is ~ 4-5 times too small, but with the increased density, perhaps smaller input current (because the electrons may recycle a few dozen of times) and thus effectively provide a higher electron density. Those are the two of the three (density is the other) contributors to the fusion rate in glow discharge fusors. If the Polywell works; this messy, cheap approach may allow for confermation of predictions, though at admittedly much lower levels than an electromagnet vertual cathode Polywell.

EG: A typical good naked glow discharge fusor might produce 1,000,000 neutrons/s at 15 microns pressure and 50,000 Volts and 20 mA of current.
IF the same grid was surrounded by permanent ring magnets with radial poles, and the yield was 100,000 neutrons/s at 5,000 volts and 10 mA and 50 microns pressure, it would be impressive. If a grounded or anode wire loop was placed just ahead of each magnet (mostly shielded by the magnetic fields) and the performance was increased further, it would be even more impressive.

Note, this speculation does not take into account the ion losses. In a gridded fusor, the ions are lost after ~ 10-20 passes due to collisions with the grid. This would still occur, but creation of new ions should be facilitated by the increased electron density within this little brother to a Wiffleball. There are plenty of neutrals to act as fuel. How this would effect the contained electrons and their energy is uncertain.

Dan Tibbets

[chrismb]

If you ran a central grid in the middle of this, then what you should get [IMHA - in my humble analysis] is the same as you would get for a fusor without the magnets, but you would be drawing less current because you are mitigating the electron currents to the outside.

I put it up on fusor.net somewhere. Mag fields need only be very modest to arrest the egregious electron losses. I reckon a mere 70 Gauss will keep the electrons from doing their worst, without loosing too much ion directionality.

In my device, I have installed a total of 182 cm^3 worth of N42, which provides a 0.12T field across a 4 cm volume depth with a 5 cm working radius. In my current set up, for ease of implementation they are within the vacuum vessel. This'd be practically impossible at amateur level as that field represents around 500W power to generate that field, whereas with permanent magnets that field is there, day in day out at no power input cost whatsoever.

You can spread or focus permanent magnet flux relatively easily, if you have the geometry that can make use of it. As mine is cyclotron (planar) like, it is very easy to use permanent magnets. The complexity of Polywell means that it is essentially impossible to do so and, as your experiment so graphically demonstrates, the electrons feed straight into the edge fields of the magnets.

The advantage of a planar system is clear - forget "superconducting" when you can use "permanent"!

[D Tibbets]

For location, your close in my instance, but I think you are off ~ 1/4 the circumfrence of Earth in ChrisMB's case.

You mention spinning electrons. I get the impression you are thinking of electrons traveling around a torus in a magnetically coupled plasma (like a Tokamak). That is completely different from what is occurring in a Polywell. In a Polywell the electrons are (ideally)oscillating back and forth from the near center to the edge of a spherical space limited by the Wiffleball border. Within this space there is no or little magnetic field(it is canceled by the competing motions of the charged particles).

Using a wire or solid surface coated with boron would represent beam- target fusion. I believe this is how the fusion crossections of various elements have been determined, but as a net power fusion approach, it is completely worthless. For net energy gains, you have to have some mechanism to conserve the energy of the particles through thousands of non fusion collisions, before a fusion collision might occur. Beam- target fusion is a one shot deal.

P-B11 fusion could be done in the better amateur fusors, if a boron gas was introduced with hydrogen. The problems is handling the boron gas compounds(purified boron 11 isotopic gas compounds would be best, and handling this gas would be dangerous), and detecting the presence of fusions. Neutrons from D-D fusion, while difficult to detect, are easy to measure because the detector can be set up outside of the vacuum chamber. With the alpha particles produced from P-B11 fusion, detection is easy, but measurement is difficult, because alpha particles will not penetrate outside of the vacuum vessel. The detector has to be inside the harsh environment inside the vessel. Not impossible to do by several methods, but much more challenging.

Diamagnetic materials might deflect an electron some, but I suspect this is a minor effect and would not affect the electron motion much at the tens of thousands of eV that the electrons are carrying at the bottom of their potential well (the Wiffleball border near the magnets). Also, if the paramagnetic material deflects the electrons some,it also, repels the magnetic field from the magnets. I think this might decrease the magnetic shielding that is needed . I am assuming that any diamagnetic gains would be trivial compared to the penalties incurred by disrupting the massive magnetic fields of the magnets. But, I don't know much about this effect. I have seen bismuth mentioned several times in the context of fusion. I think it has something to do with it's very high Z. Bussard mentioned it in passing when he talked about the CNO fusion cycle rescueing his 'Bussard Ramscoop' concept in an interview. Other than this mention, I have not found any reference to bismuth and the CNO fusion process.

The potential well already exists for the electrons. The trapping Wiffleball border represents the lowest potential energy barrier. The adiabatic recoil from this border is what turns the electrons around for the next pass towards the center*. Some material barrier would be a loss mechanism, if the partical hit it. As mentioned above, if something impedes a magnetic field on one side, it also effects a magnetic field on the opposite side. At best, I would suspect that the net effect would be neutral, but I doubt that equality could be reached. Due to thermodynamics, the gyroradus of the energetic electrons, etc. I would expect a net energy loss.

* Either the electron recoils back towards the center (limited focus due to the convex curvature of the magnetic field lines and transverse thermalization), or they become trapped on a field line. The dynamics of these competing processes (within the confinement lifetimes of the electrons) is unknown to me.

Dan Tibbets

[MSimon]

I hate to be a spoil sport but you still have field lines going into the faces of the magnets. Even with ring magnets.

Now if you just "want to see what happens" and don't care about particle losses. Have fun.

[D Tibbets]

True, permanent magnets are hungry sinks for charged particles, with the possible limiting condition where the poles of hollow cylindrical magnets are on the lateral sides, so that the poles are slightly past the tightest portions of the cusps. I speculate, that even without recirculation, this may approach the electron confinement described by Bussard as cusp confinement (~5-50 electron passes). This would exceed the electron confinement of simple central cathode gridded fusors by that same amount. As mentioned by Chrismb, the ion losses to the central cathode grid would be unchanged.

What intrigues me is if this leaky magnetic confinement would supplement a ETW setup- no cathode grid, but a peripheral anode grid just in front of the magnets (with electrons supplied from e-guns). This grid would be shielded to an extent. This would work like a poorly functioning Polywell. It would be relative cheap to build and operate (if adequate vacuums could be obtained).
At the least it should limit the power supply requirements (voltage and/ or current needed to reach measurable fusion). At best it might allow for measurements of actual Wiffleball formation and function at levels similar to WB4, with significantly boosted neutron production, etc. As the permanent magnets would be limited to ~ 1 Tesla fields, the output would never reach break even. But the neutron output might reach several orders of magnitude above WB4, possibly to levels where a viable commercial neutron source might be obtainable (I'm thinking of neutron production in the 10-100 billion per second range). Of course, if a Wiffleball is desired, even with WB4 confinement quality, the current input would need to be considerable. Considering the density, volume, B- field strength, and assumed confinement efficiency, the value can be calculated. With ~ 1 Tesla B-fields in a small machine, compared to WB4, the current requirements may be relaxed. I am assuming the increased B-field in the permanent magnet setup (~ 10,000 Gauss vs 1300 Gauss in WB4) would more than compensate for the presumed modest shortfall in cusp electron confinement that I would hope for if the B field strengths were the same.

[EDIT] One unknown in the above speculation is how much recirculation occurred in WB4 (with it's large doghouse interconnects, with no separation of the magnets, and square cross sections). Apparently WB6 was ~ 10 times better in this regard, but that does not indicate where the WB4 baseline recirculation factor was. Did it improve confinement by 1.1
X, or by 10 X, or by 100X, or by....

Dan Tibbets

[hanelyp]

Any prospects for a design where the electric grid and magnets are separate? An electric grid inside the magnets so the electrons are slowed before they reach the magnets, magnetic field reaching in to shield the grid.

[D Tibbets]

That is what I envision. Look at the second set of pictures (with the magnets arranged in a ring). There is a central cathode, and a wire loop just inside the magnets which was grounded in this instance. A positive charge could be applied here, though I'm uncertain what is going on with the relationship between the grids. Without the magnets I have had cathode voltages of several thousands of volts (limited by the vacuum levels I can achieve), but the peripheral anode voltages seem to be limited to a few hundred volts. I've only used a NST (Neon Sign Transformer) for the anode. The voltage drop with modest currents in these transformers may be playing a role, but I'm uncertain why the current would be higher in the anode (of course in a pure system the electron and ion flow/ current between the cathode and anode would be balanced) but I don't know why the positive anode voltage is limited compared to the cathode. I really need to do some current measurements and some head scratching to figure it out.

Dan Tibbets

[ortholithiation]

Isn't Boron 11 80% abundant. You could make a borane gas B2D6 as a gas, as long as you are under vacum its really no big deal. It would react with water to generate D2 and boric acid in the pump oil or you could trap it in a cold trap.

Just purge the chamber with loads of N2 or Ar afterwards to dilute.

[butterfingers]

Where did you get the wire cathode?

[butterfingers]

Also is it a cold cathode or hot cathode? Does it matter which one is used?

[butterfingers]

must it also be wire?

[D Tibbets]

Assuming your questions were directed to me, the central wire cathode is typical of those used in amateur IEC Fusors. The cathode is a cold wire initially and will produce electrons through field effects (it is fed a current of a few 10's of mA at ~ 500 to 10,000 Volts. Once electrons are emmitted air in the chamber (at ~ 1/10,000 atmosphere) is ionized into a plasma. The wire then heats up due to the high speed positively charged ions hitting it.
It is a simple way to create a luminous plasma within the magnets. To be a step closer to a real Polywell, the central wire cathode would need to be replaced by a virtual cathode through methods discussed in Bussard's literature. I have done this (poorly)in some configurations, but the light is dimmer, and it is more difficult. In any case the permanent magnets do not allow for much confinement - cusps run into the surfaces of the magnets.

It provides some visual effects that represents somewhat how a Polywell confines a plasma and the shape of a noncompressed "Wiffleball", but not much else.

Dan Tibbets