Microwave Ion source

[Doug Coulter]

It came to my attention I'd neglected to put a link here to my microwave ion source, which I'd documented on the old web page. So here it is.

This is one of the larger successes around here -- it's performed like a champ for 100's of hours without a glitch, till the wire to the puller extraction electrode came loose inside the tank, which will be easy to fix the next time I have to door open. I don't yet know how well it will work with run-of-the-dumpster microwave oven tubes, because the first one I tried worked great, and no reason to mess with it further, as in, this one works, and no one has offered to pay me to make them one. I used the lowest power tube in the junk pile, which seems like it was a good idea.

ECR (electron cyclotron resonance) ion sources have some cool features. No metal electrodes in the "nasty zone" so no sputtering. Since it really is an electron cyclotron, once it's started you can take it ridiculously low in pressure and it will stay lit. Even at crazy long "mean free paths" for the gas present, the electrons just keep going round and round till they hit something and ionize it. Where nearly all other ion sources make mostly diatomic ions (in this case D2+), this one makes predominately monatomic ones. And mine at least will run down to the e-5 millibar range easily, once lit at somewhat higher pressures.

For what it's worth, after a long good fusor run, we sometimes see neutrons coming from the ion source with everything else turned off. And you can turn them on and off with the source power -- perhaps some of the fusion byproducts (T and 3He) are easy enough to fuse for that to happen. All I know is, it happens and only after a long fusor run when the various byproducts have built up. I can also see them on my mass spectrometer, as I pump the system back down through the pressure range it will work in.

The nice thing about the 2.45 ghz frequency is the magnetic field needed to get to resonance is big, but not too big. In other words, other stray fields (like the earth's) don't mess it up, but you can still easily generate the field needed with good NeFeB magnets you can get fairly cheaply. For lower frequencies, the cavity gets a little too big to be practical, and the stray field issue is larger.

We recently acquired some ~8 ghz klystrons I may try this with too, but they are pretty low power devices (fraction of a watt) and may not do for this. If so, we may use them for plasma diagnostics instead.

[Joe Jarski]

Doug, I've been studying your ion source and was wondering if there is any possibility of driving multiple ECR chambers off of a single magnetron and cavity. For instance, if you spaced the quartz tubes at the proper multiple of the 1/4 wavelength and end cap from the tuning screw? I think tuning would be nearly impossible along with some other practical issues, but from a theoretical standpoint?? Or am I way off on my Mwave theory?

[Doug Coulter]

There's plenty of power there it seems, but the whole cavity is just barely over 1/4 wave -- the center rod is tuned to that with the tubing in there (the thing is really a 1/4 wave piece of hard-line coax shorted at one end, and fed in the middle -- super-high volts at the end of the rod). You might be able to do a power splitter and drive a few cavities. What I find with the particular magnetron I'm using is that it mode-hops when I crank it up above about 50w output, which makes the source blink and not smooth -- but you only need 20-30w anyway for it to work nicely. But these things are sooooo cheep...if you need a few, heck even new ovens are about $30. Dumpster ovens a little less...

It would be a significant improvement if the endcap that retains the tubing were a screen instead, and the whole thing in the tank instead of having to send the ions down a skinny tube.
You'd have to insulate all the metal at the HV end of the cavity, or it'd sputter, and make the RF feedline vacuum tight to do that. You'd get 100's of times more net ions that way. I plan to try it when I get some time for it. And unless you got really fancy with motors, tuning would be a "break vacuum and open the door" kind of thing, might be frustrating at first. It would be especially cool if the end screen were floated DC off the cavity (bypass caps) and set to pull ions out actively. You'd be in the hundred ma range then -- more than too much, but also very controllable.

Sigh, just another thing to add to the infinitely long priority list of things to do! It might be moving up the list soon, as the opportunity to make more fusors has been happening --we just got another batch of turbos....

[Joe Jarski]

Thanks. That gives me a few ideas to work on if/when I need multi-barreled source. I went back and salvaged a 500W magnetron from a microwave that I scavenged the MOT from a while back. Hopefully that'll be small enough to get started with one similar to yours. More projects...

[Doug Coulter]

Yes, more projects indeed. People have done this in a waveguide vs a cavity, but the advantage of a cavity is that I can get more volts/meter at less power using the more "discrete" tech, for which we're about at the frequency limit at 2.45 GHz. I couldn't really make the quartz tube bigger, as that would spoil the cavity and leak too many microwaves. If you were willing to run much higher power density, you could have glows at all the voltage peaks along the waveguide, and presumably you could do the ECR magnets in discrete chunks along the length, and even absorb much of the energy (to cut standing waves and SWR that can fry the tube) by the end of the waveguide, though all the implementations I've seen need a dummy load at the end. Since each high voltage node would have less and less power as you go down the guide (they'd be half a wave apart), each source of ions would be a little less good....you'd have to do this in a insulator-lined waveguide so as not to sputter metal as well. It might have to be (ugh) ceramic to avoid the reduction to metal that happens with silica, but I must say in the existing source it seems electron bombardment of the quartz is cancelling the ion bombardment in terms of reducing it to metallic silicon, which doesn't happen in the fusor main tank -- it becomes shiny metal and failures then occur in that place, due to the DC preferentially attracting ions to the insulators (which are never perfect, and the worse they get, the worse the effect).

Complex enough for you? Some seemingly simple designs are actually pretty elegant, when all this is taken into account.

[Joe Jarski]

Most electronics are still way over my head - especially RF stuff. It's good to know that I'm on track and picking up bits and pieces as I go. Just getting a single microwave ion source working good would be a major accomplishment for me. Even though I somewhat understand the idea, there's some sort of voodoo magic that happens in the cavity/waveguide.

Maybe it's time I crack open my books by Terman!

[Doug Coulter]

Terman is the man on this (though some radio am handbooks are good too). We're making a piece of coax, 1/4 wave, which if shorted at one end, has infinite impedance at the other end. So the volts at the end of the tuning rod are "infinite". Not really but very high. We couple energy into this someplace lower down, like the tap on an autotransformer. The hardline to the magnetron is also critical as an impedance matcher from the tube to the load...it's not exactly a quarter wave (closer to half wave) and adjustable to tune out some reactance in the ends.

Yes, getting one to work is a good place to begin. Note you really want quartz tubing for this -- pyrex is lossy enough to act like a load and it gets hot and wastes power.

[Joe Jarski]

I've been reading Terman's books on transmission lines, waveguides & resonant cavities and have a lot better understanding of what is going on in the cavity. The one place that I got stuck was the size of the cavity that you used vs the calculated size of a resonant cavity. That one took a while until I realized that it's not a resonant cavity, but just a shorted transmission line. You had said that several times in all of the different write ups - it just took me a while to sort it all out. The magic is starting to disappear. Now on to the power supply.

[Doug Coulter]

Just to add a little magic back in -- see this. This is his re-entrant cavity, Pg 265, fig 131 i.
Equation is pn pg 268, eq 150. It's the one I implemented in the code on that thread, with 1.5 fudge (see discussion in book) added in.

But yes, this was a transmission line tuned circuit for an express reason -- I could get higher volts/meter off the end of that line for given power in than I could in almost any cavity design.

Now for the one on this thread -- what you want is kind of a constant current supply (the magnetron almost acts like a zener diode), and in this case I controlled the current via a much smaller than usual series cap in the doubler (about 1/10 nominal), and got the "constant" by the added diode and cap combo for output filtering. And then ran the whole thing on a variac. Not many milli-amps to the tube...There's a sweet spot on the variac where more makes it blink and mode hop (but is easier to light off) above it, and no operation below it. In my rig that's about 91v nominal in, and the transformer is happy and not saturated and buzzing and heating itself. I'd about bet that some series R after the output cap wouldn't hurt to keeping the AC ripple off the uwave output power. Or for really old-school, a big fat choke (10's of henries, current is real low here).

[Joe Jarski]

Yeah, I was looking at the re-entrant cavities while I was reading and thinking that they had a nice E-field for this sort of thing. I thought that might be a good experiment down the road once I had better grasp on this one. I got a chuckle when I saw your post on the new ion source because I was thinking the same thing. I thought the E-beam coupling to re-entrant cavities on pg 271 was an interesting method of excitation too.