Yes, a little theory here, but backed up by experiment. The video embedded here shows some data from two runs with different grids, all else identical insofar as I could. I'm displaying kV and mA on the hoizontal axes, neutrons detected/min on the vertical axis, and pressure as color. You'll note pressure was mostly different on the "new gen" grid - lower, since for whatever reason, it'll run lower for the same current, probably because it's a bit longer and the wires are smaller so they make a bigger local field gradient, but for what I'm showing it doesn't actually matter too much - we're still pretty close in pressure on an absolute scale (or about the same mean free path).
What this shows, is that yes, like the kids say "crank it to 11" to get more - that's not where the Q is happening, not hardly. There can be a number of possible reasons for that, one being things like space-charge defocusing, but there can be others as well. Here's the vid:
http://www.youtube.com/watch?v=-58gdiLNIqc
I show both raw (though converted to real-world units) and Q plots for both runs. The "old" twisted grid run was done with the exact same setup as the run we did the day before HEAS 2013, only we did this one the day after - with Joe Jarski here to help and observe (more than a little out of his way home from Richmond; thanks for coming, Joe ). The other screen shows the stuff from the new grid, during breakin bakeout - I tend to let the "base pressure" which is pretty high for awhile after opening the tank, get to say, e-6 mbar (which only takes 10 minutes or so), then put in some D and run - this "cleans up" stuff a lot faster than merely heating the tank with my heaters (gets it hotter, and D reacts with a lot of things to make stuff the pumps can pull out easier). So that was the "break in" run, or perhaps the first of a few, for that grid. Not quite happy yet with the adjustment vs the length of the sidearm, it might be better pulled in a little further, we'll see - I already pulled it in 1/2" since it was a bit longer than the old one (an advantage of my FT design is that you can do that and rotate things, often without even leaking). But it might work better back in the more-uniform field, I'll try that too.
For reference, here's a pic of the older, twisted grid. Other than that, the differences are, the rods (tungsten TIG rods, because they are cheap, and very straight) are larger - .040" in the picture, where in the new grid they are .020", and the new grid is about 1/2" longer than the old.
This one does not throw a focused beam out of the end, or if it does, at least it's way out of focus by the time it gets to the viewport, about 15" away from the end. That could be construed as either a bug or a feature. The new, straight one, DOES toss quite an energetic beam of "something" that is repelled when I put negative volts on my ion source grid, which is halfway between the main grid and the window, but off to the right side, as I demo'd in one of the videos on my DCFusor youtube site. The bases on either one get quite hot. It's obvious why on the old one - it's thick and conductive carbon. With the new one, not as obvious why, as its thin, anodized Ti.
However, experimental details aside, with mostly the same gear (other than the grid) - I see the same stuff, in large part. And that is that Q is highest at the bottom, not the top, of the current range, and it's a big factor - 100's or a few times that.
We're doing it wrong! Have been the whole time .
So, at the very least, focus really does appear to matter - it increases the "density" of the virtual target for each beam. When the current is low, and the pressure is low, we see far higher Q for both grids - some of that could be fewer collisions with neutrals at low energies due to lower density elsewhere (less gas). I made both as accurate as I could, but that's a long way from even optical perfect, much less short wavefunction perfect - the errors are on the order of mils.
This lead me to think, OK, even if we did have some ideal world (which we very much do not) going on here, with just positive ions, that actually acquired the full eV of the power supply (no way), if we had perfect focus, how close would we be to being able to "force" them together, by some definition of "together"?
So, I ran some back of envelope numbers. No, these aren't right to a bunch of decimal places, I used old data for weights, charges, e/m etc, but to a first approximation, at least the order is right, and the first digit or two certainly correct.
Now, the generally accepted radius of a D ion is1,88e-13 cm. If one posits, as most books do, a square well for the strong force - well, it's a little larger, about 5e-13 cm. Let's use that number for the moment, as it's a nicer number, gets us closer to where we want to be. I found some good math for closest approach of two nuclei with a certain momentum, and worked that all out on the envelope, and the answer is, with zero impact factor (perfect head on hit with 50kv net - assuming 25kv/ion and head on) : 2.84e-12. So, we're never getting to that square well at all, or it's not square, or we'd have no fusion at all. I'm still working on the math for tunneling probability vs distance - but I can't find a decently understandable formulation so far - or even one that might take into account the possibility that the well isn't square. I got the math I did use for this from a "new" book BillF found and brought me (thanks!).
Due to the various calculations, where volts and velocity show up as sometimes sqrt, sometimes square, it works out that the distance of closest approach is 1/Volts times some constant for a given ion set. Meaning if we want to get to that square well, we need more volts - on the order of 50 * 2.84 * 2 kV or thereabouts - to get the wells to touch. Actually, that's a bit pessimistic, because that multiplier assumes the actual edge of one D ion gets into the square well of the other - it's a little better if you work it so you take into account the size of both "wells". But still horrible, and still making the assumption that I'm even getting half of the supply volts onto these ions, which I believe is a terrible assumption under the current conditions. It seems quite easy to have so much space charge in there - the extra electrons and such - that various shielding mechanisms are at play here (well known in the plasma biz, but the new book I have on that doesn't state what units...so I'm working on getting a better understanding so I can use the "simple" equations they did the usual cheats on - mix units to avoid having to write c or 2*pi and so on. Si, MKS, ESU, EMU tend to get mixed to make the equations "beautiful" but I need real numbers, doggone it. You know, the type engineers depend on to do a good design.
In fact, that alone easily explains why for example, a 1955 Phillips book that shows neutron borehole tube design shows that it gives about a million neuts/second per microamp, at 120kv or so. And here we are using 20,000 times as much current to get the same result in most cases. OK, so they use a DT mix, which if one is pure D and the other is pure T, gives you 100 times or so the cross section at resonance. That means we're still doing 200 times worse than a beam on target borehole tube here. Well, except, as you can see - in the case of very low current at the onsets of "lighting off", in which case, we're doing a lot better(!). Not enough to beat them yet - I saw 600 ua at about 750k neuts/second kinds of numbers.
So, we have some ways to go. Our "target" isn't as dense with D's for one thing - so we miss more. On the other hand, our target is effectively "deeper" since in beam on target, all the ions that hit the target lose energy very quickly (near the surface) due to interactions with the other atoms in the target, usually some metal that stops D really well. In a beam on beam situation, you don't lose energy as quickly with each scattering, so the target, while less dense (but getting better as focus improves), the loss per scattering event is lower, so that makes up for the lower density somewhat.
Given all this, it's sort of amazing your average fusor works at all! For one thing, we're getting a lot less than the applied field onto our ions. For another, we are working with a very un-dense target, though I have other evidence that with our level of focus, we are actually making the transition from molecular to viscous flow - we have "compression", in effect. In an early run, we tried a cylinder grid, closed off at the end, but stuck a ceramic tubing, 1/8" ID by an inch long into a hole in that closed end. Ions come streaming out of that pipe...something is pushing them out despite them not seeing the tank field at the other end of that pipe from inside the grid - viscous flow inside, molecular outside (lower pressure and longer mean free path).
So, where to next? It only took a few years and some hundreds of thousands of bucks to get to the point I could write this with some confidence. There are other things that need to be measured if I can figure out how. There are other things to be calculated once I get the math straight and the units right - that charge sheilding that happens in most plasmas. Of course, this isn't most plasmas, not at all - else we'd see a uniform kind of glow, not beams, so...more to learn. I'd love to actually know the voltage gradient between the shell and the grid - where's the drop - but at present, even with a wiggle stick etc, I can't just stick a probe in there, as the presense of the probe, and the wire leading back to a ft so I can read what's on the probe - is enough to make the fusor not work normally anymore. I need to think of another diagnostic for that, or make something almost magical - it's not like insulation that will hold off 50kv with no leakage is made flexible so I can just have a "dot" sensor in there with no other side effects - oh, and the insulation would have to be more or less infinitely thin on top. That's not gonna fly with what I can put together here, but there's probably another way...it's a topic of thought experiments at present.
Doppler shift of the light given off by recombining ions in the beams? Who knows? It's a sure thing *I* don't, for the moment. It's also a fairly sure thing I'll find a way at some point.
We do know there's a large excess of negative charge in the tank, even far from the main "action" as determined by the faraday probes I have put in there. But at this point, we don't know how much is electrons and how much is "whatever" flavor of D- or D2-, or what. There is evidence that at least some of the D is picking up enough electrons to be negatively charged, as we see fusion at the tank walls - we've made a "tandem" accelerator, at least for some of them. From what I measure, about half the total fusion is happening at the walls where the beam hits. That's one reason I favor the hornyak detector - it's tiny and doesn't smear out the spatial information like a huge moderator for a 1/V type detector does, so I was able to see that effect at all. Tyler C saw a similar effect using a BTI to get spatial resolution. We both thought at the time we might be seeing neutron beaming, but I don't think so now - it's just locality of the detector and the fusion.
More when I sort more stuff out. Feel free to chime in with any ideas. I am also thinking abou spin as a way to increase the effective size or probability of fusion within a certain range. Spin is conserved - I believe Pauli mentioned that back in the '30s or so. As it turns out, in a D nucleous, both spins are the same (proton and neutron alighed). That gives spin 1 for a D (each nucleon has spin +/-1/2). He has no spin - so you'd have to have two D's with opposite spin to make that He, for example (He has spin zero).
Or not - does a photon come off, and carry away some spin? How many and how much per? Dont' know. The 16.7 MEv Q of that reaction, the rare one, is so high, I'm informed that the He is unstable and decays into one of the two more common reactions after being formed - T +p or 3He + n, in order to carry off all that energy, and make the spins come out right - which is obviously easier with more particles (and one possible explanation of why TD has such a higher cross section than DD).
Lots to think about here.