Proposed new uWave Ion source

High frequency, antennas

Proposed new uWave Ion source

Postby Doug Coulter » Sun Dec 12, 2010 3:38 pm

With the advent of the cheap little turbos my personal goal of making a nice, portable demo fusor (that of course pushes the boundaries in other ways) is looking more and more possible.
As we've done our other research, we find that we do better the lower the pressure we can run at, down to the point of needing an ion source to make it go at all -- well off the Pashen's law light off point, and at a point where you pretty much need to provide all the ions some other way, not just a trickle. Nothing has yet beat my other ion source design (adapted from RevSciIns), but it's on the wimpy side for a main source. So, I've been thinking on how to do all of the above. Since this is primarily an RF design exercise, it landed here -- we already know that we want an RF cyclotron source, that discussion is elsewhere.

For this sort of thing, what you want is an E field that a lot of volts/meter (concentrated) for the amount of power input, crossed with a static H field at the cyclotron strength for electrons at the frequency in question. For 2.45 ghz that's in the 980 gauss range, not too difficult for modern permanent magnets, even over a fairly large space. But getting a high E field to power ratio needs something more than a waveguide. In the previous design we used a quarter wave tuned line to get high voltage at one end. In this design, I'm going to use a re-entrant cavity for that purpose, and build the thing right into the ISO nipple between pump and cylindrical tank, leaving the tank innards as pristine as possible. The idea here is to have it close, straight shot for the ions to get into the tank easily, and simple (at least in appearance ;) )

Here's a rough drawing.
Sketch -- graph blocks are 1/4", nearly at scale.

From the outside it will just look like a fat place in the vacuum line pipe, with a couple of FT's brazed in.

You can think of a cavity of this type as being tuned by the capacity between the screens in the middle, across the inductor that is a torus around the outside. Energy is coupled in with a loop from a microwave source (in this case, an oven magnetron at 2.45 ghz). The E field is concentrated between the screens here, vertically. A static magnetic field on the order of 980 gauss will be applied horizontally with permanent magnets to get electron cyclotron resonance over some of the area between the screens -- it's that trick that makes this work down to very low pressures, otherwise known as long mean free paths, as the electrons just go round and round until they finally hit something and ionize it.

Tuning is a sensitive function of inter-screen distance (eg capacity). So at least one of them has to be a sliding fit and adjustable. If it's mounted in the end of a slightly smaller piece of Cu tubing that is insulated from the 1" tubing by mica, it's at the right potential for RF, yet can have DC applied to it to "push" the ions out into the main tank area. Various other schemes could work as as well, or be added to promote ions into the tank through the final, grounded screen that keeps the in-fusor DC field shape uniform (eg no big dip where the hole would be).

Here's some code I wrote that does the equations from Terman. The book claims the original equation will be off by factor between 1.25 and 1.75, so I built a 1.5 factor in -- 2 pi becomes 3 pi.
Code: Select all
use strict;
#use warnings; # can't becuse of a slungy trick to have defaults

my $minorrad = 1.27; # assume one inch inner pipe
my $majorrad = 3.81; # assume 3 inch outer pipe
my $majorheight;
my $minorheight;
my $temp;
my $twopi = 3.1415927 * 3; # fudge factor of 1.5 added in already
my $wavelength; # the answer

# get inputs
print ("Supply me with some info on the cavity- all in cm.\n");
print "\nInput minor radius ($minorrad) -->";
chomp ($temp = <STDIN>);
if ($temp + 0 > 0){$minorrad = $temp};
print "Input major radius ($majorrad) -->";
chomp ($temp = <STDIN>);
if ($temp + 0 > 0){$majorrad = $temp};
print "input minor height -->";
chomp ($minorheight = <STDIN>);
print "Input major height -->";
chomp ($majorheight = <STDIN>);
$majorheight /= 2.0;

$wavelength = (($majorheight * $minorrad*$minorrad)/$minorheight) * log($majorrad/$minorrad);
$wavelength = sqrt($wavelength);
$wavelength *= $twopi;

print "\nResulting wavelength = $wavelength\n";

# done, exit

To run this on a windows machine, you have to have Perl (free from ActiveState) and probably change the "shebang" first line to point to where perl lives on your system. Or just read the docs for the one you get. They're recently broken the Tk interface for Gui's on this, so my gui-perl won't fly on windows without getting the Tk module from the perl repositories. This one is just a command line app, so no sweat this time.

Here's how I see this when editing it in linux/gedit, a lot more clearly.

Screen shot of simple program and its action.

All this is in centimeters, and for reference, we are trying to tune for 12.24 cm wavelength here (eg 2.45 ghz).

I have some defaults you can override in the program, but the assumptions are using 3" Cu pipe for the outer part of the re-entrant, and 1" for the inner size, though it's possible to make this tune with 1.5" inner size too -- just pickier. I believe that a 1" hole will let enough pumping happen for a small (6" diameter by about 1 foot long) fusor, if it has no leaks and few extra surface area contributing "warts" welded into the tank. I plan to use glass ends, covered by screen wire, one drilled to let the HV feedthrough be glued in. Viton o rings for the glass to metal seal to handle the tempco force monster there. Might even use an ignition system to run it pulsed mode, there's some good stuff out there for hot-rods that makes 50kv and more at decent currents.

This should let me toss it in the back of the truck, even run it off the truck's AC inverter....or put it someplace far from me and run unattended for long periods to activate stuff. It will obviously be a lot easier to shield the X rays from than the big tank in the lab as well.

We will feed this with a telescoping hardline I'll make in the shop. We'll use a retractable antenna for the center conductor (if I can find one still! Kind of obsolete these days) and sliding Cu telescoping outer pipe for the ground part, held with a clamp. I will taper this line (have to anyway to make it telescope) so that it also acts like an impedance stepup from the nominal 75 ohm output of the magnetron to a few hundred ohms Z at the coupling loop in the cavity. Hpefully, I can also get some fine tuning that way by coupling some reactance into the loop when the match isn't perfect, and still not have it arc or fry the maggie tube via SWR. Something tells me this might just need more RF power than we got in the last one, we may not be able to go CW with this, but for a demo, it may not matter (and we might use one of my HF switchers and get close enough to CW anyway -- this will ring for awhile when the power goes away between pulses and the ions coming out will also time-smear some). With the little turbo, I can use the same gas control I use in the big system, and let in D in the foreline while the turbo is spinning slowly. So all the "warts" will be on that one ISO nipple, or between the pumps.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
Posts: 3024
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Proposed new uWave Ion source - Cavity parts

Postby Doug Coulter » Wed Dec 15, 2010 3:45 pm

Got a little shop time in and made these:
Cavity parts

These are all the major parts, except the screens across the 1" pipe pieces. I cut the original nipple shorter, and was planning to use the piece left to go cavity to tank, but borked it, so I'll just use a little bit longer Cu pipe there.

This will be brazed, all but one joint, which will be the top plate soft soldered to the 3" ring, so I can take it far enough apart to adjust the input coupling loop off the type N fitting shown in the picture. I'll try to get the RF coming in as high in voltage as I can (via odd number of quarter waves and tapered line) but obviously there's a limit there with arcing, so the rest of the stepup will have to be the cavity itself -- yes, it's one turn coupled to one turn, but since the coupling loop will be smaller....well, it works out. The impedance seen will obviously vary all over the map as gas conditions change, but that mostly amounts to a resistive load. I'll make a yoke to mount magnets outside of this to get the crossed field needed to make it work down to really super low pressures. The calculated gap is a big wide for best RF performance/voltage multiplication, but you have to leave some room for the electrons to whirl -- they won't get going all that fast before they hit a grid as is.

I'll probably silver plate this after all -- I couldn't bring myself to dice up the nice thick copper sheet I have (could probably buy a car with it), so I used thinner but still insanely expensive brass for the top and bottom. Since it won't take part in the discharge, it seems OK to use this -- I won't be sputtering zinc off it anyway. But the material for the screens....should be high temp, electrically conductive, and hard to sputter. I have tungsten, tantalum and nichrome around, as well as titainium. Manganese would maybe do. I'm leaning towards tantalum if I can join the stuff with a spot welder or equivalent, as it doesn't sputter too much. Tungsten + traces of water or oxygen is not good -- as a last resort, I can chrome plate whatever I use, as that's supposed to be hard to sputter too. I'll just have to run the thing and see what works, so nothing new there. I'm going to attempt to get it basically tuned in plain air, outside the vacuum system first, using extra input RF power as needed to get to the light off point if I can, as that will make it go a lot quicker. I should probably also drill a small hole in this someplace to see the internals so I can tell if it's working, and how the glow distributes for fine tuning. I will also add a hole for one of our pirani sensors -- a grain of wheat bulb. These go lower than normal piriani or thermocouple gages, to the point they are still useful at fusor type pressures, though that's about the end of the useful range. It doesn't hurt that they read much higher on light gasses in this case.

I decided to make the top screen adjustable and the one that gets an extraction field on it. I'll do this by making a plug out of telescoping tubing (actually, I'll probably machine that out of rod stock) that's a little undersize and insulated from the main 1" pipe by mica. That can have a wire to a negative supply, or just depend on electrons hitting it from the fusor grid to get a charge there...most of the "current" in a fusor is electrons, due to secondary emission from the fusor grid after all, so there may be enough current to do that -- in the big fusor there's plenty showing up on a faraday probe after all. This way, I can reach into the tank and push/pull it down and up for tuning, no need to take things far apart.

So, that's plan A. IF we don't find some SS pipe to make a mini fusor from, we do have a 6" by 18" flanged copper pipe we could use with glass ends. I've had good luck making o ring sealed windows for this sort of thing, and can hole glass pretty easily, so I may solve some HV feedhtrough problems by just making the thing it comes in through a big chunk of glass, rather than a tight fitting grounded piece of metal. Of course, from experience, we know the glass won't live forever and will take some pains to protect it. Luckily, it's cheap anyway.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
Posts: 3024
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Proposed new uWave Ion source

Postby chrismb » Wed Dec 15, 2010 4:46 pm

Is there some more reading on this type of ion source, Doug? I have to confess that I do not see how it constructively resonates like a magnetron... or does it work more like a virtual anode source?

(PS... not wanting to drift the thread, but have you considered UV lasers to stimulate ionisation directly around the grid?)
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Proposed new uWave Ion source

Postby Doug Coulter » Wed Dec 15, 2010 6:43 pm

Second part first -- yes, UV might help, but with long mean free paths, maybe not as much as you'd hope -- after all the electrons streaming off the grid won't keep the thing lit below some pressure, and I want to go lower than that with this. you know there is already a ton of vacuum UV in a fusor, right? It's enough to sunburn you through a Pyrex window which "theoretically" stops that stuff, and generate ozone around the windows. There's just too little chance of it hitting the right thing and having just the right energy, unless I suppose if you could tune for that.
Where I'd find a 13.6 ev laser or source I don't know....that's way above even vacuum UV, into the low X rays, sort of.

I will drive this resonant cavity with a magnetron, it's not a magnetron itself, just a tuned circuit with a handy shape for my use here. Looking at my sectional drawing above, you can see the outer part of a cavity is a one turn inductor, which is resonated with the capacitor formed by the reentrant smaller pipes with screens on the end. No magic here. I will add the crossed H field the same way I did on the other one. Since the electron cyclotron gets huge long electron paths around and around, they have more chance to hit something and ionize it per electron than the ones off the grid that have a 3" max flight path, then hit the walls. When mean free path gets into the 3" range, no ionization happens, or not enough to keep it going without some help, which would be this thing (assuming it works, of course).
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
Posts: 3024
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Return to RF

Who is online

Users browsing this forum: No registered users and 1 guest