Debye Length

[TallDave]

Yeah, I'm not sure what the deal is with voltage. Art pointed out a while back Bussard wasn't maximizing output, which suggests there is some unnamed limiting factor, perhaps arcing or increased transport or greater temp spreading.

The gain's variance with B is so strong under Bussard's loss equation that I tend to think all his slack was in there. Bigger doesn't help you nearly as much; you only gain by r**3/r**2, or linearly, with size, versus B**4/B**.25 for the magnets.

Of course, once you get to around 12-13T, you can't do much but make it bigger, which is why ITER costs $20B.

[MSimon]

Yeah, I'm not sure what the deal is with voltage. Art pointed out a while back Bussard wasn't maximizing output, which suggests there is some unnamed limiting factor, perhaps arcing or increased transport or greater temp spreading.

The gain's variance with B is so strong under Bussard's loss equation that I tend to think all his slack was in there. Bigger doesn't help you nearly as much; you only gain by r**3/r**2, or linearly, with size, versus B**4/B**.25 for the magnets.

Of course, once you get to around 12-13T, you can't do much but make it bigger, which is why ITER costs $20B.

The ideal voltage from a gain perspective is the first resonance peak. Which comes at 50 KV nominal drive (about 65 KV actual). It reduces power density by 10X compared to operating at 200 KV (actual) at the cross section peak. Gain goes up by 2.5X at the lower voltage

When working with SCs smaller is better. B goes up with lower size. The smaller the size the greater the power density. Doc B was thinking Cu magnets. With those you do gain from larger size.

But all the above assumes a coil intercept of 20%. With the higher fields channeling the alphas I think an intercept of up to 50% may be possible. Or even 70% once we learn something.

[D Tibbets]

deleted

[D Tibbets]

...

But all the above assumes a coil intercept of 20%. With the higher fields channeling the alphas I think an intercept of up to 50% may be possible. Or even 70% once we learn something.

I'm guessing that by increasing the 'intercept', you are referring to thicker magrid casings. Since the magnetic field strength drop is dependent on the distance from the magnetic field generating wires, having superconductors spread out more in larger casings would allow lower proximal fields to deliver stronger fields to the cusps- especially the central cusps. Also, of course more room for shielding and cooling. This would help copper coils also (room for more windings).

Dan Tibbets

[hanelyp]

The ideal voltage from a gain perspective is the first resonance peak. Which comes at 50 KV nominal drive (about 65 KV actual). It reduces power density by 10X compared to operating at 200 KV (actual) at the cross section peak. Gain goes up by 2.5X at the lower voltage

A narrow pB10 resonance peak like the first would be fine if the polywell were a 1 dimensional colliding beam machine. Since we're dealing with collisions at a range of angles, and effective energies, the much broader second resonant peak looks better.

[MSimon]

The ideal voltage from a gain perspective is the first resonance peak. Which comes at 50 KV nominal drive (about 65 KV actual). It reduces power density by 10X compared to operating at 200 KV (actual) at the cross section peak. Gain goes up by 2.5X at the lower voltage

A narrow pB10 resonance peak like the first would be fine if the polywell were a 1 dimensional colliding beam machine. Since we're dealing with collisions at a range of angles, and effective energies, the much broader second resonant peak looks better.

Very good point. However, it may not be a show stopper. Experiments will need to be done.

[BenTC]

I've been reading around trying to get my head around debye screening and how it relates to the Polywell. I came across the following which I thought might apply. However, how is the Polywell categorised in terms of collisional/collisionless?

While shielding in collisional plasmas obeys the standard Debye result, shielding in collisionless plasmas is far more complex than commonly believed. For example, a one‐dimensional (highly magnetized), immobile‐ion plasma can, in some circumstances, anti‐shield a positive test charge; i.e. the plasma becomes more positive in the vicinity of the test charge. When shielding does occur, it results from electrons dynamically trapped in the neighborhood of the test charge. A new theory of collisionless (Dynamic) shielding in one, two and three dimensions is presented here, and is in excellent agreement with experiments in pure electron plasmas. Because the distribution functions found in Dynamic shielding are highly non‐Maxwellian in the non‐linear regime, collisionless Dynamic shielding can be substantially less efficacious than collisional Debye shielding

http://socrates.berkeley.edu/~fajans/pu ... byePoP.PDF

[D Tibbets]

The ideal voltage from a gain perspective is the first resonance peak. Which comes at 50 KV nominal drive (about 65 KV actual). It reduces power density by 10X compared to operating at 200 KV (actual) at the cross section peak. Gain goes up by 2.5X at the lower voltage

A narrow pB10 resonance peak like the first would be fine if the polywell were a 1 dimensional colliding beam machine. Since we're dealing with collisions at a range of angles, and effective energies, the much broader second resonant peak looks better.

The Polywell in essence is a series of colliding beam machines organized in a spherical geometry. Of course this is an ideal and will never be reached. But, if the energy of the particles are narrow enough, and the convergence is great enough, this ideal could be approached to the extent that this narrow crossection peak could be taken advantage of. Perhaps enough (eg:30%) of the ions will fall in this range, so that their contribution plus the contribution from surrounding energies could be enough to make efforts in this range worthwhile. The net yield from this 'relaxed' energy distribution around the resonance peak' would be enough to make it profitable. When increased losses from drive energy and bremsstrulung at higher energies is factored in, it might result in a larger machine, but it also may be the only way to exceed breakeven enough to be usable.

Dan Tibbets

[TallDave]

I've been reading around trying to get my head around debye screening and how it relates to the Polywell. I came across the following which I thought might apply. However, how is the Polywell categorised in terms of collisional/collisionless?

While shielding in collisional plasmas obeys the standard Debye result, shielding in collisionless plasmas is far more complex than commonly believed. For example, a one‐dimensional (highly magnetized), immobile‐ion plasma can, in some circumstances, anti‐shield a positive test charge; i.e. the plasma becomes more positive in the vicinity of the test charge. When shielding does occur, it results from electrons dynamically trapped in the neighborhood of the test charge. A new theory of collisionless (Dynamic) shielding in one, two and three dimensions is presented here, and is in excellent agreement with experiments in pure electron plasmas. Because the distribution functions found in Dynamic shielding are highly non‐Maxwellian in the non‐linear regime, collisionless Dynamic shielding can be substantially less efficacious than collisional Debye shielding

http://socrates.berkeley.edu/~fajans/pu ... byePoP.PDF

Great find Ben!

It sounds like they mean collisional in the sense of Maxwellian, which PWs aren't. So this paper seems to confirm what a lot of us had suspected: you can't apply standard Debye screening in a PW.

It's interesting they verified this in Penning traps.

I'll have to dig through more thoroughly, it might be possible to calculate a Debye length at WB-100 conditions from this. (I suspect Rick and others have already done so.)