I've been thinking about grid designs as incremental (not revolutionary) ways to improve fusor Q somewhat, as I have found that doing so even in the worst "static" modes, that doing that also helps the higher Q pulsed modes work better anyway, and it's easy to test. I think that electrons released due to fast ion impacts with just about any conductive grid material are most of the losses, along with mis-focus, so improving both grid accuracy and reducing electron emission should help, I believe. Here is what I currently have in mind, and since it's not really hard to try, I'll be trying it fairly soon. Sigh, things never move as quickly as hoped...so the definition of "soon" is a variable here (and everywhere else, it seems).
First of all, I've found it's not insanely difficult to machine pure graphite, which has one of the lower secondary emission constants of conductive things, at least when hit by electrons at kilovolt levels -- I can't find data on responses to fast ions anywhere, so there's an assumption that is untested so far. The one pictured has a mistake in it, but I'm going to try it in the tank as my second grid anyway -- you never know what you might find out doing things like that. The mistake is an imperfect symmetry of the vane spacings. Sometimes that helps things, so who knows? I will, soon enough.
Shown next to that are some 1/16" pure alumina rods I got from Accuratus (I have ones half that diameter on backorder, they've got to be hard to make!). The idea here would be to align these along the vanes on the outside around the grid, and cover the inactive ends with larger alumina tubing so there's simply less stuff for an ion to hit head on that is carrying a big excess of electrons due to the power supply input. I'd think that the ions will still see plenty of attractive field from the sides and inside ends of the vanes so it should still work. This would make most electrons released be set free inside the grid, where they might actually do some good, instead of instantly flying to the tank walls to make heat and X rays, or so I'd presume. They'll want out of there, of course, or -- if there's enough positive charge in there, maybe not! Seems worth a try, at any rate under the meme of "try the easy stuff first", so that's the plan.
To make this, I first make another graphite grid, presumably this time without the angular spacing errors (but more on that later*), only first I cut a 1/16" deep and wide groove along the length where each vane is, before I machine out the inter-vane stuff. I could then lay the insulating rods in the grooves, and hold the whole mess together with a couple of tight fitting alumina rings at the ends, which would also get rid of the ions that hit those to at least some extent. The thickness of alumina I have won't hold off the full supply voltage, or so I think, but it may still insulate some, and not so readily supply excess electrons when collided with. Or that's the hope.
*
Various endeavors have found that uneven symmetry is a good thing. It could be perfect symmetry is poison for various reasons. For example, my best main grid right now has a bit of twist to it, due to a machining error (I'm good at making mistakes!). Could it be that's an impetus to get whatever minor amount of recirculation we get -- helps organize the flow in some way?
Could be. An example is Eric Lerner's DPF rig. At first, other researchers had great difficulties replicating his results. He found that it only took the earth's stray field to give just enough impetus to things to make it work, and other workers had been much more scrupulous about making the setup field-free (until the big pulse, that is). When he redesigned to put just a slight twist on his apparatus, and others also did, they all started getting the same results...
I know that as a gunsmith, that the very best barrels (which is the determinate in accuracy, or a main one) don't have an even number of lands and grooves, and one pays a high premium for ones that have 5, rather than 4 or 6. The theory there is that the bullet is deformed less when two lands aren't directly opposite. And it works out in practice. I've built several rifles, by the way, that are accurate enough, at scale, that if they fired protons, I could hit one with another very, very reliably -- and frankly, protons ought to be easier, since they are all identical, unlike bullets (and brass and powder charges), and can be made to all go the same speed at the same time (unlike bullets). There's no wind or barrel vibration problems with charged particles, comparatively. For reference, these super accurate bench-rest guns shoot into a quarter inch (all the same hole -- it just gets a little ragged after a lot of shots) at 400-500 yards on a calm day. And this is with "barrel whip" that if the bullets were to come out at opposite peaks of that vibration would be thrown several feet one way or the other at a mere 100 yards, which is easy to prove by simply putting a finger on the barrel during a shot and seeing how much that throws off the shot. If you imagine a garden hose with water pressure suddendly applied and the end whipping around, you'd be close (and I have some cool movies and simulations done by our defense department that show this).
Could it be that having a vane or rod opposite an opening on the other side is a good thing? Seems no one has tried it, or at any rate, reported on it, so it's something else to try. One could build a model where this helps focus at the interaction zone, so it seems worth a go even if it results in no recirculation. In other words, an odd number of holes or vanes might be a good thing, not a bad one, especially assuming there is any good effect from electrons released on the *inside* of the grid volume. I think there is simply no way to be sure without a try, so I will.
I believe that in addition to the math, which is more useful in hindsight than foresight sometimes (think the three body problem in gravitation, as we have the analog going here), getting that "feel" or gut level understanding about how an apparently simple system can produce emergent behavior is key to being good at this game. Or, well, it has been for me, and I've had a few surprises along those lines already.
I bought a better degree-ing setup for my machines, so maybe I won't make those mistakes so badly, we'll see how that works out when I get some more graphite to chew on (very soon, and Jon's will be in the mail to him soonest). At this point, we don't actually know if it's a good move to work with that stuff -- seems reasonable, but it's unproved, I've only used it for end-caps with W rods, so far. One thing that does seem reasonable is that I can at least theoretically, make these much more accurate than grids made of metal I can machine (can't really plan on cutting tungsten, though I might try titanium at some point if I get a better micro-mill for the lathe toolpost put together).
My composite grid suffers from fit and finish issues -- can't make a perfectly aligned interference fit hole for the rod ends at this point, maybe not ever, where when machining out of a solid chunk of it (inaccurately called billet, as Jerry has pointed out) I can get it a whole lot better, to near the accuracy of my machine ways if I don't flub it up, which is more than ten times better than anything I've yet put in the tank and tried, so far. I have to repeat that so far, grid accuracy has been a super big effect once I started getting anywhere close to good. The difference between huge and medium error wasn't much, but things got better fast going from medium to small, and in this case small is still pretty big compared to what I can theoretically do.
So it seems the game might be worth the candle on that one.