I've been getting some email asking me to kind of lay out how to build a fusor for a newbie, and rather than answer them all one by one, I'm starting a thread here I can send them to.
This is kind of like the classic old story of elephant stew -- first, go get the elephant, technique left to the reader. It's one of those "conceptually trivial" but actually not so much kinds of things.
You can describe a fusor construction loosely in about one sentence.
Create a controlled low presssure atmosphere of fuel, put a grid in it, and apply high voltage to that grid.
Sounds pretty easy. But each part of that sentence can take a bit of doing, to say the least.
At least in my case, the elephant was the vacuum system. You need to be able to get down to some microns (one sucessful guy reports running at 16 microns), or somewhere in the range of 1-2 e-2 millibars in your chamber, while having good fuel purity. Although it's been done, barely, with a 2 stage mech pump, you'll not get very far on that "purity" that way, and will never get outstanding results. Sorry, but vacuum is a whole technology of it's own to learn about. Almost everyone needs a fore pump of some sort, either a diffusion pump (cheap, but hassle to use) or a turbo. If your pump gets the chamber down to 5 microns running, what that really means is the chamber is outgassing enough to stall the pump at 5 microns, so if you then seal it off, the pressure in it will rise again. Zero pumps achieve nameplate ratings in use, they are rated with a blocked off intake and after a bake and a long pump -- no chamber outgassing is taken into account for that number. To get it clean, think about getting to say, 1/100th the pressure you're going to run at (eg perhaps a pump rated 1000x better than your target pressure), and you'll be in the ballpark there. Your chamber should be something like stainless steel, or other material that won't out-gas a lot, and can be heated at least somewhat to speed that process up. You'll need a way to get into it to play with grids and so on, and you'll really want a window to see in, and a way to get the HV into it, along with perhaps some other feedthroughs for probes to measure things like field distributions, temperatures and so on.
Gas control is incredibly harder than anyone just starting out realizes. A tiny controlled flow of gas is very hard to get to, as we all find out at some point. I am the only one I know of who has managed full sealed-off operation, having an especially clean and tight setup that stays pumped down continuously between runs -- months at a time and it gets pretty clean in there and stays that way. Everyone else winds up doing a flow through system, and balancing the input flow rate with the outflow is a real challenge to say the least. You'll need a good regulator for the fuel, a decent valve, and probably either an orifice or a piece of tiny capillary tubing to get any sort of control of that, and even with all that it's like flying a helicopter blindfolded, upside down, and will take full attention and both hands to keep in balance. Some tricks are discussed elsewhere on the forums. One can also work with throttling the pumping system in various ways, but the bottom line is you're trying to match a supply with a leak and keep a precise level the whole time -- it's inherently a hard thing to do and always requires some active control.
High voltage might be the "tiger" flavoring for this elephant stew. It can be easy or hard, depending on what you can get your hands on, or build yourself. Many of the commercial supplies are pretty good (I admit to being a Spellman fan), but you'll have to provide some sort of ballast impedance (I use resistors and sometimes inductance there). With no ballast, peak currents from the capacitors in the supply can get to be so high a gas discharge develops into an arc at low voltages, so you must limit peak currents. While there's always a temptation to think just a little more would be better, in terms of power in watts, it's not so true -- we have tried it all. Things melt! There is nothing much to carry heat away in a vacuum, so things that don't have a conductive path to the outside are going to get really hot, and you may as well realize that up front. Not only that, at typical fusor dimensions, anything much over 500 watts input is going to get the tank walls so hot there will be other problems and issues, so that's about the limit for most fusors that are normal sizes. For starters, you'd want something on the order of 40kv at 10 ma available to get into a good region of operation, and if I were to increase any of that, it would be to go up in volts, not current, as my tests here show a quicker rise in output that way than just going with higher current at lower voltages. The lowest we've ever seen countable neutrons at is around 20kv, and at 50kv we're in the millions (same current either way). Part of that is it seems fusors run better at lower pressures, which comes naturally when you try to limit current at the higher voltages...fewer collisions of fast ions on slow neutral background gas that way, more of the input goes into doing something productive. The limit there is Paschen's law - below some pressure it won't light off at any voltage you have unless you also build a good separate ion source. Note the horizontal axis is pressure times distance there...
Getting HV into a fusor can be a real trick, as mentioned elsewhere on these forums. The standard HV feedthrough design has a rod "floating" in the vacuum. But it's not a vacuum, and that long uninsiulated stalk creates a long path for Paschen's law to get going in ways you won't like. Look around here for some ideas on what I've done (and anyone else please add stuff) to overcome some of the problems and expenses there. You can try a commercial porcelain feed through, but don't be surprised if you destroy it fairly quickly due to hot ions bashing it and reducing it chemically to a non insulator, there's some arcane tweaking to be done there. Due to Paschens law, which we work on the left side of the plot, things that won't arc or discharge over short distances will over longer ones, kind of counter-intuitive, but there it is, as you will find out.
We have tried a few different grid geometries here. Suffice it to say that you are, like it or not, building an electrostatic lens when you do this, and you'll need to make it so it creates a good focus where you want it. If you're of the sphere religion, then you need to have a spherical outer wall too or there's no prayer of doing that, and if cylinder grid, then cylinder tank.
Either have some theoretical problems. You cannot get equal spacing on a sphere, period, you'd need a design as complex as a buckyball to even get reasonably close. With a cylinder, neglecting end effects, you can at least have equispacing of the rods, otherwise known as "lens elements". This is geometrically impossible with a sphere, but many neglect that thousands years old knowledge because the idea is "pretty".
To stay alive, and make progress, you'll need some measuring tools. A good pressure gage of course, though after awhile you can adjust by how it looks -- the thing itself gives the information, and is more sensitive to small changes in pressure than most gages! Likely your entire operating range will be inside a minor division on a gage, and the gage is mainly useful to tell you which direction you're way off in. You will need a for real geiger counter or other way to measure the X rays you're going to produce so you don't get hurt. Windows are especially bad leaks of these. We see two classes at least here -- one is copious X rays at the power supply voltage, the other is very hot gammas from a rare DD reaction pathway that doesn't make neutrons, but goes straight to He and a gamma -- 16 megavolts or so. If you have energy discrimination, this can be used to know how much fusion you're making. Of course, you'll want meters on your power supply too.
You will want a neutron detector of some sort, of course. These can be hard to scrounge (very expensive new), but worth it. Types are 3He tubes, BF3 tubes, B10 tubes, BTI bubble detectors, and activation of for example, silver in a moderator, that you count with your geiger counter after a run. The last is cheapest by far, most reliable (insensitive to EMI and all sorts of other things that mess up electronic detectors), never wears out, and gives a way to compare results fairly well between labs. I'm getting a thread ready to describe how I do that for all. The downside of activation measurements is that you don't get any information in real time which is kind of needed to get adjusted correctly, so I use both sorts of thing here. I'll say this though -- when the silver activation and the tubes don't scale together, it's the tube I suspect, not the silver. At any rate, I use the electronic detectors for real time feedback, and the silver data to know if this was a better or worse run than the others.
This is a more daunting task than many of the newbies I work with imagine -- they seem to think they can just take one from this, one from that, all off the shelf but cheap, and immediately do better than (ah the hubris of youth) the guys who have been doing this for years. Perhaps because they have some "hot new idea", not realizing that so did everyone else, and we don't talk enough about how those all failed when we tried (note I have a whole forum up here for that for that reason, but it's hard to get people to admit failure so as to save others).
That's pretty unlikely. It took me a couple of years and more than one fusor setup to implement that key sentence above, and I substituted money and off the shelf new stuff for a lot of the scrounging, but still wound up having to do a lot of fishing around on ebay and elsewhere and making various adaptations to get to "first neutrons". And a lot of hours as well. After that, and having gotten a feel for it all, improvements came very quickly, but...some things you just have to find out hands-on, and most of the things you think you can cheap out on or get away with -- nope, you can't, it all has to be right *at the same time* which implies that all the pieces have to have even higher individual reliability, in the manner that multiplying probabilities will show. For example, what power do you have to raise 80% to to get under 50% chance of it all working? It's not a big number, something under 3. So three subsystems that are all 80% reliable means the thing will rarely work as a whole...Once you start counting each piece here (all gaskets aren't one piece for example) it's a big number so you need to have what engineers call "nines", or things that have a lot of nines in the reliability numbers if you want the thing to "just work". For example, 99.99% isn't a bad target for each thing, and more nines is better.
Details on the individual subsystems needed are all over this set of forums, it's far too much for one post, this is merely an overview. Knowledge as provided here will help you get going and waste the least money. As my dad once told me, the reason poor people stay poor is they buy crappy junk, and have to keep buying more when it breaks, or spend all too much on incremental upgrades till they finally wind up buying the one thing they should have bought in the first place -- all that other money and time was nearly 100% wasted.
At the bottom line, while the information here on this site can and will help a lot, counting (or depending) on getting it right on your first try is not a real smart move. Any real budget has to have some fudge in it for the unexpected, because in truth, it's only truly unexpected if you don't think much. And this is more science than technology, it's not a canned forumla you can just loosely copy, forget the details, and expect to get great results. In other words, it has more the flavor of research than building a kit from plans. You can do the latter, but the results will never be as good as what you'll be able to get with some iterations once you learn the basics.
Anyone who has run a household budget with some success eventually knows this. People get sick, cars break down, interruptions occur (not all bad but if you had other plans for that time...) and if you don't expect and plan for a bit of that, you're just a bad planner, or so vain you just don't think it can happen to you, but it will anyway! You'll have to plan for a little bit of failure or your plan is a bad one, it's "jus zat simple", to coin a phrase.