Inverter Microwave for Ion Source RF

For the design and construction details of ion guns, necessary for more advanced designs and lower vacuums.
lutzhoffman
Posts: 188
Joined: Tue Dec 08, 2009 1:59 am
Real name:

Inverter Microwave for Ion Source RF

Post by lutzhoffman » Mon Dec 21, 2009 8:49 pm

Hello:

In the interest of building a microwave ion source, I have been looking for an old microwave to gut for RF parts. Doug Coulter is currently building exactly such an ion source, and his work in this area is brilliant to say the least. He is much more knowledgeable on the subject than I am, and he is kind enough to share his work. Between Carl, and Doug's work, quality ion sources will soon be a simple build for the amateur. In my opinion this is a fairly profound event, which may lead to many great discoveries in the future. As far as nuclear physics on an amateur level goes, ion sources ARE a very big deal, and perhaps the most difficult single component home build.

Originally I planned on using Carl's design, but I soon discovered that it is very difficult to obtain the required set of RF components, which are small enough to fit inside of the HV terminal of my project, even with the aid of a small amplifier, like the ones from Communications Concepts.

This is where Doug's idea of using microwaves comes in. With a microwave based ion source the nearly complete set of RF components, can be had at the local Walmart for under $50! Better yet they fit inside of a small space, with simple control circuits. If my ion source were at ground potential then I think I would go with Carl's design simply because it is the proven, and traditional way to go, with few unknowns, or potential issues. So in a nutshell I like both approaches very much, but I want to try Doug's design on this project.

I recently scored a working Panasonic Inverter Microwave oven, rated at1350W, at the local dump. To my delight it is in perfect working order, the interior has some burn marks, and a broken glass plate, so the owners must have "assumed" it was toast, or that it was to expensive to fix.

I very carefully removed all of the innards, being careful not to mess anything up. This went very well, I was even able to remove the control panel leaving it attached. After a short time I was able to get the magnetron to work outside of the oven (While observing RF safety measures) just fine. By leaving the factory control circuit attached, I have full control over the output, both in terms of power, and time. My feeling here is that instead of trying to engineer a new control circuit, it would be very easy to just wire an extension from the control board to the magnetron. I will just have to make a translation table like: 1 Pound Auto Defrost = 100uA Beam x 10min, this could be handy : )

What I cannot figure out is the properties of the microwave output from this Panasonic assembly. Traditional tranny based microwaves produce a pulsed output, via the HV diode / capacitor circuit. Looking at the Panasonic inverter based design, I notice that the input 120V 60Hz, goes into a FW rectifier and is first converted into DC. The DC then goes into a circuit with what looks like a fairly solid MOSFET, or IGBT, then to an inductor / small capacitors, and to the HF ferrite transformer, then finally the magnetron. It could be a resonant system. The circuit however lacks anything like a substantial HV capacitor which has me wondering about the nature of the RF output?

Some promotional literature on the net seems to imply that the output is CW? By referencing to a traditional microwaves "pulsed output" being inferior. My question is: Is the output of this inverter microwave pulsed, or is it CW? Or if it is pulsed, is it pulsed at very fast rates like in the KHz range? I do not own a scope so I cannot just figure it out this way

I could not find a clear answer about this question when I researched on the net, which led me to writing this post. For an ion source CW is normally preferred, but when I think about it the efficiency of ion sources normally goes up with increased RF power. Thus a fast enough pulsed RF output could be an ideal situation. The higher peak RF energy increasing efficiency, while the HF pulsing rate is fast enough to enable easy power measurement, and monitoring. Personally I do not want to have a 60-120Hz ion source output, but 10 or 20 KHz would be just fine, and maybe even preferred to CW : )

Can anyone shed some light on this, by explaining the nature of the RF output from an inverter microwave?

Thank You.....Lutz

User avatar
Doug Coulter
Posts: 1312
Joined: Sun May 27, 2007 7:18 pm
Real name: Doug Coulter
Location: Floyd, VA, USA
Contact:

Re: Inverter Microwave for Ion Source RF

Post by Doug Coulter » Mon Dec 21, 2009 10:01 pm

Well, you speak of the devil, and here I am.

I own two of those ovens myself, they are fantastic for cooking.

No, they aren't CW, they pulse at 30khz or so. But to reduce power, they reduce the pulse height/width at that frequency. This looks enough like CW to the food to make no difference, compared to the old on for a few seconds, off for a few seconds older machines used to control power.

Getting a maggie to go really CW is kind of a trick, and one I guess I should start writing up for others.
My basic idea is far from new, I copied the basic RF design from a rev-sci-ins article from way back there, where they used a medical diathermy source (so they didn't have to figure that part out at all) and were mainly interested in cavity design. I just copied theirs more or less with some slight improvements of my own, and added the ECR magnets they forgot to put in there for better performance, and indeed this is better than they claimed by a large factor. If I can believe my mass spectrometer it makes mostly atomic ions, not D2+. I even see that with the mass spec fairly far away from it in the tank -- I see some D2+ too, but that would be about the amount from some hitting the walls and recombining on the way to the spectrometer.

I am debating whether to make a real long thread entry here on "how to make a magnetron that was never meant to run CW do it anyway" but I'm having second thoughts as it would be pretty long.
You're not going to get there with either the panasonic maggie (far too big) or that power supply (completely unsuited for this work on several levels). I am not sure either of the power supplies I've done this with would easily float up to HV, mine are operating near ground. They wouldn't go into a HV terminal easily at HV unless there's one heck of a fine isolation transformer, or a shaft drive/generator kind of thing around, these things take some real power input, about 80 or 90 watts to run.

Lutz, if you wanna email me, I can send the sloppy version, I want the one I put up here to be pretty good for beginners to follow.

But the key is this:

Find a magnetron tube about the *lowest* power you can find -- the panasonic one is a monster. We just picked a small one from dumpster microwave ovens, but they're not expensive at the microwave repair stores either.
You may need to try more than one anyway, as they mode-hop at low/medium powers, in and out of the resonance of the high Q cavity sometimes - a small one does this less as there's a lot of stored energy in the cavity to help keep it from doing that. You must run them very low power CW, as these have no facility to handle electron buildups and eventually you burn out the ends when the electrons spiral up and down to them, or in some, just build up in the tube cavity and eat RF power and stop all oscillations. A little tilt on the mag field stops that one, easy, and usually is there by "made sloppy and cheap" for free.

You then need a "constant current" well filtered DC supply, at pretty low current, 10-20ma max. I've made two, one with standard parts from an oven (but I used a transformer at higher volts than required for this tube so I could afford to lose some in the filtering) and one that uses my H bridge switcher. They both work fine. You will need a fan on the maggie, the filament power alone will get them above max spec temperature without one. That's around 33 watts there. Then you put in maybe another 20-50 watts DC. Above that, things fail quick, and it's more than you need anyway.

It's a balancing act running this thing right at were it will oscillate at all, not mode hop, and not burn up, but once you find that point, it works every time you flip the switch, very nice -- I've run it for many hours (maybe 100 or more, and many off on cycles), no problems. I am using a variac in mine with the standard xfrmr, set at about 89 volts....higher and it mode hops and becomes useless (and burns up) and lower it just won't oscillate.

To make a supply with a standard MOT you need to use a small series cap, on the order of 1/10 the one they normally supply (I am using a .1uf and it's still too big, about half that would be better), a series R and another diode/cap (2 uf used here) on the output to get decent DC. I am using about 2k series R there and have a neon bulb with it's own ballast across that as a simple current monitor to see if the thing is drawing some current and a rough indication how much. On peaks the drop across 2k/10w R is plenty to light it.

Kind of surprised the board let me put up this 807kb file, but I copied and improved their cavity number 5 -- just a few helpful changes and builder tweaks to keep the arcing down and make it out of easy to get parts, mainly. Like I said, not many watts input are needed. I made a yoke and with magnets from amazing magnets to do the ECR part on this. I adjusted the field with a hall effect meter for the rignt number in the middle, it goes up on both sides of that, but seems fine and works fantastic -- I get the thing lit off, and it stays lit to e-6 mbar levels -- very nice if you want to run low pressures and not need differential pumping to keep the rest of your apparatus down there for long mean free path.

Quartz tubing is a must for this. It works, barely, with pyrex, but needs higher powers, and basically it's the pyrex getting very hot and needing a fan. The cavity (1" ID Cu pipe from the hardware store) gets warm, but only warm. I use thin wall 1/2" OD quartz for this. I let the gas in via a capillarly tubing in the fitting on the outside the tank end.

First picture is the thing in action, you can see it lighting up the gas. The picture isn't good enough to show that it's kind of pinched and expands away from the cavity. See fan on maggie tube, that's absolutely required. This is the business end of the tank, and I put that plexiglass shield on there to keep arcs from the main feedthrough down.

The next pic is the power supply I'm using. It's a standard MOT, with a wall wart glued on top for fan power. I have kept the original series cap on there in case I someday want a lot of power, but for now, it's not hooked up. You can see the end of the white series cap I am using, near the to-220 diodes I series-ed up to get high enough voltage for this (they are 1.5kv 6a diodes, takes 5 for this use). The big cap on the back is the main filter cap, on which the charge is lethal, which is why it's buried in my rack -- safety, and no I'm not pulling it out of a working system for pictures!


Hope this helps. Don't tell everyone about the copyright violation that paper represents, just get it quick before we have to take it down.

Oh, further issues. I made the endcap on my lathe, and charge that with a positive voltage, and a aluminum tube was turned to fit into the quartz on the tank end, and extend some beyond this so any sputtering there doesn't make the quartz conductive. These are fed from a CCFL inverter with two half wave volt doublers for + and - polarity output, which is controlled by an adjustable LM 317 on the input of the CCFL inverter. Works like a champ, and lets you control the amount of ions injected very nicely and very fast.

In my recent record runs, I had to turn this off. The lm317 puts out enough with the adj terminal grounded to put in too many ions (about +/- 300 volts), and enough just drifted in to make that mode work. For other modes I use about +/- 3 kv or so, with about a 33k resistor on both leads for arc protection issues. Those burn up before the rest does, kind of a fuse.
Attachments
UWavepower.jpg
UWavepower.jpg (60.8 KiB) Viewed 4180 times
uWaveIonSrc.jpg
uWaveIonSrc.jpg (61.94 KiB) Viewed 4180 times
Microwave Discharge Cavities Operating at 2450MHz.pdf
(807.56 KiB) Downloaded 180 times
Why guess when you can know? Measure!

lutzhoffman
Posts: 188
Joined: Tue Dec 08, 2009 1:59 am
Real name:

Re: Inverter Microwave for Ion Source RF

Post by lutzhoffman » Tue Dec 22, 2009 1:11 am

Great Reply, Thanks Doug

Ok, There is one thing that I just am not getting, sorry, I am not trying to be difficult:

If you feed the microwave ion source with 30Khz pulsed microwaves from an inverter microwave unit, instead of the normal CW microwave RF, would it not still function? Instead simply producing an ion beam which pulses at 30KHz?

I am asking because Sulfur lamps for example are fed with pulsed microwaves, and the discharge is not extinguished in between pulses or anything. This has me thinking that if a Sulfur lamp; Which is nothing but a hollow quartz ball filled with low pressure Argon with some elemental Sulfur, works just fine on pulsed microwaves. An ion source is basically the same thing, just with an added magnetic field, and an extraction electrode. Maybe a RF driven sulfur lamp could even be converted into an ion source, by drilling a gas/extraction hole, and adding some magnets? The trouble is the sulfur lamps are hard to find in the first place : ) This brings me to another idea, just for fun fill some quartz balls with various elements, and energize with RF : ) A Rubidium fill focusing on a Nd:YAG rod would be interesting. Anyway back to the topic....

My thinking here is why not power the ion source with pulsed microwave RF, if it is fast enough to sustain the discharge? If an ion beam pulses at say 10-30KHz, then the beam current could be averaged out via a capacitor in the faraday cup circuit for an accurate reading. In other words it would not matter to me if my particle beam pulses at 30KHz, as long as I get an high enough average beam current.

Maybe the ion source could be used both ways: With CW RF for precision work, and with pulsed microwave RF for brute force high beam current work?

Thank you for your thoughts, and for explaining the concepts involved.

Aloha.....Lutz : )

User avatar
Doug Coulter
Posts: 1312
Joined: Sun May 27, 2007 7:18 pm
Real name: Doug Coulter
Location: Floyd, VA, USA
Contact:

Re: Inverter Microwave for Ion Source RF

Post by Doug Coulter » Tue Dec 22, 2009 3:35 am

Yes, you'll get a pulsed beam with a 30khz component to it. It won't quite match the drive, as some ions get going fast enough to make more by collisions, the electrons will continue their cyclotron motion for awhile, so there's a tail, the length of which depends on a lot of things -- how much gas, volume, and so forth. If you use enough extraction field, that may do some keep-alive as well.

In my case, I couldn't tolerate the resulting noises as I am studying time-dynamics in the fusor just now, and didn't need an additional signal to have to deal with, or have affect the thing I'm looking at.
All my phenomena don't mind the 2.45 ghz, that's too fast for ions to do much. And this thing btw makes so little uwave leakage a microwave leakage detector that does register the oven I have, just sits on zero anywhere around it. Nice.

Sulfur lamps are designed to be easy to light off (most lamps are ;~). This means that they can light off on every cycle easy -- and they are dumping real significant power into those anyway. They'd be using in the few torr region of gas to promote this, just like merc vapor bulbs do and for the same reason. Low pressure is a relative term, they are not low in terms of what you need for accelerator runs, but *many* orders of magnitude higher than that.

For just making ions, you don't need and can't use high power -- things melt and so on.
Note the fan they put on the one in the paper, for a 75 watt RF source. Mine doesn't need that and makes plenty of ions, orders of magnitude more than you'll have the power to accelerate. Or I do, and I have kilowatts for that at much lower volts than your accelerator will need to be interesting.
My source puts out at least 10 ma as configured. That's more than plenty for an accelerator that would typically run in the small uA region at most.

If you try this at the half power minimum the panasonic gets in near CW, before it reverts to the same old on-off for a couple seconds each, your stuff is going to go up in smoke in a rather spectacular fashion. Even the quartz will melt. That's ~5 times the max power used in the paper, btw.

The beauty of CW operation is that once lit, it will stay lit down to ridiculous low pressures, where over-simple armchair theory says "no way, mean free path far too long". If you let the thing drop, it's hard to start again. Mine seems to take about 3-4 e-5 mbar or more to start, but will run way below that, but only CW or very fast at any rate -- I tried 70khz and that wasn't fast enough to keep the ions hot until the next pulse. It had to be in a gas condition allowing restart every time, more pressure than I think you can tolerate. For a fusor, it's probably fine to pulse if it's fast -- it will start every cycle at those pressures, I am just looking at something special and don't want the distraction just now.

You are going to find that to get high average beam current you'll need something more like DC, not a high peak to average ratio of pulses.
The focus will change with instantaneous current due to space charge effects, and there will be a maximum you can keep focused to your specification, whatever that is. You don't want high energy particles to almost make it to the target, but hit the tube walls, for example. Ideal focus potential depends on current in ANY beam device, from CRT's on up. I've acquired a heck of a lot of literature on all this, and tested a lot of it -- they are not telling lies. Too bad I can't upload the nearly 10 gb we've gotten hold of in fusor/fusion/beam technology papers and related stuff, it's quite the nice library.

Remember, during off times, gas is still flowing into your accelerator, just not ionized. You're going to have enough troubles keeping that pressure down low enough to get a mean free path you need without letting in gas when it's not even getting ionized, and that's going to be kinda hard to pulse too at those rates. The big boys all use another vacuum system (called differential pumping which has nothing to do with diffusion pumps) pumping out a space between orifices between the source and the accelerator -- the ions go right through the holes when in focus, the other gas bounces around and gets removed, mostly. If you're worried about what you'll need to put in the HV terminal....well, a complete additional vac system sounds kind of troublesome to me. And no, a long quartz tube from ground won't do it as the gas in there will be highly easy to ionize at that pressure....which is why it was AT that pressure to begin with.

In fact, I did all this due to my Scottish repugnance to the idea of another vac system being needed, and found a way to avoid that need, even for pretty low tank pressures. At that, it's quite cheap by comparison.

To be honest, it looks like the source Seltzmann is using might be the better one for most fusors, it's certainly simpler to make. I doubt it lasts as long without maintenance as this one, just because this one is pretty phenomenal in that regard -- it's just never failed in many many runtime hours, and no hint it's degrading in any way at all. THAT is cool indeed.
Why guess when you can know? Measure!

User avatar
Doug Coulter
Posts: 1312
Joined: Sun May 27, 2007 7:18 pm
Real name: Doug Coulter
Location: Floyd, VA, USA
Contact:

Re: Inverter Microwave for Ion Source RF

Post by Doug Coulter » Tue Dec 22, 2009 5:08 pm

In answer to a question Roman asked about the transfer of power from the tube to the cavity:

It's not a wave guide, it's a short (about 1/4 wave) piece of homebrew hardline coax. I don't want to tear this down just now for pictures, but I can describe it fairly well. A wave guide would have been quite large for this frequency (a couple inches across) and had too low an E field in it, with a quarter wave cavity we can concentrate the E field far higher right where we want it. So this is all kind of a standard low frequency RF design, just real small. The cavity itself amounts to a 1/4 wave line with one end shorted and is on the large side for this frequency. Trying to go larger diameter would allow other monkey modes to occur in it, rather than the longitudinal one we want.

The top of the magnetron is the usual antenna post, surrounded by the RF ground gasket weaving.
I found a Cu pipe adapter that fit that gasket, and was big enough to hold the antenna post without clearances being too close. This adapter is brazed into a piece of sheet metal so the big end protrudes about 1/8" inch, and bolting the sheet metal down presses it into the grounded gasket.

Into the top is inserted a piece of 1/2" Cu tubing. Here in the US we call it pipe if we're talking about specifying the inside diameter and tubing if outside. Go figure, it's not a system I invented, just have to live with it.. This is all stuff you can get at Lowes or the local equivalent. I get most things like this at McMaster. Sad that I had to buy a 25 foot roll of 1/2" tubing to get 2 inches, but them's the breaks.
A stock of various tubing sizes is sure handy around the lab anyway.

The inner conductor is a piece of 1/8" tubing with about a 1/16" ID. One end has a washer brazed on it, which is then soldered to the end of the "antenna" on the tube, using a special flux for SS and nickel from McMaster again. The other end is drilled and tapped for a 4-40 screw, which is the adjustable length (for coupling to the cavity) part. This wound up more or less flush with the cavity wall, and the end was "rounded" or "melted" to eliminate sharp corners that caused arcing. Trying to tune this while in operation was a joke -- anything you put in the cavity burns up and mis tunes it, so we had to tune with power off, try, retune, etc. This coupling adjustment doesn't seem very critical at all, since we're running *way* below the tube rated power, some reflected power is fine as long as nothing arcs and there's enough forward power coupled. This enters the cavity sort of tangentially like in the paper (the picture there is all I had to go by myself). Tempting to try on the next one to place it closer to the "ground" end of the cavity for lower impedance there, but works as it is, near the hot end just above the quartz tubing hole so it's coupling almost to the tip of the tuning 1/4 wave stub.

This tuning process was helped a lot by having a DC discharge in the ioniser quartz tubing -- which eventually became the pushme-pullyou ion extractor stuff. You then always have some light and can tune for "most smoke" more easily. Once it's all tuned, it stays that way really well, I've never re adjusted that parameter for any set of conditions -- it works fine, it ain't broke, and I ain't fixin it no more.

The cavity is about 2" long inside. It's a piece of 1" pipe (eg one inch ID). I made the top piece which is threaded for a 1/4-20 brass tuning stub screw, and that's pretty thick in and outside the cavity so as to have plenty of thread length, it's about 1/2" or a little thicker, turned out of 1.25" Cu billet stock. I had to grind off the threads and round the end of the screw near the quartz tube to cut down arcing there. I also put a spring under the screw head, as this tuning is very twitchy indeed -- the entire range might be 20 degrees of turn on that screw.

Again, resonance was found by tuning this while there was already a discharge going on in the quartz tubing -- it's easy to tell when you get it right that way. Every now and then, a tiny jiggle on this screw helps the thing get going in high vacuum, other than that I never touch it anymore. That's good as in use it's very near the HV and I'd rather not "go there" when in operation.

The cavity bottom is a pipe cap for that size, that I shortened before grooving it and the cavity tubing for the quartz to pass through -- you can take this off without breaking vacuum. I drilled and tapped both that cap for 4-40 and the cavity tubing at the top for the same screw size so I can take all this apart. I should probably just solder the top on now, as that joint is a high current node in this design and more conductivity is better, it's probably the main loss I now have.

Allow me to stress again that you NEED quartz tubing. Yes, it will kind of work with pyrex, at about double the input RF power -- dicey due to arcs, and all that additional power goes into heating the pyrex, not good. As is, the cavity doesn't need a fan at all, barely gets hot, never hot enough to burn you. Yes, it's very clean inside, I did the usual anal prep there, and may silver plate it for fun. It works fine as is, however -- I'll plate it if it starts corroding in there.

A key is finding the "right" magnetron. this is a smaller one from a junkyard oven. Some magnetrons do a thing we RF guys call mode hopping (I guess laser guys use the same term, and it's the same thing). There are usually 8 cavities in a magnetron, and never are they all just the same, so sometimes one or another will "take control" and it will then go at that frequency. If that's not the frequency this very high Q cavity happens to be tuned for, it stops working. Some of this is controlled by the amount of power input, so finding one that doesn't mode hop much at the desired power level is key. Having it be a smaller tube helps here too as the amount of stored energy in the cavity can help keep it from mode hopping better -- we're adding a ninth resonator, a better one than those in the tube, even under load.

It seems to be a luck factor, but luckily magnetrons are cheap even bought new. The key, again is getting a *small* one. We don't need the big power here at all, too much of a good thing is bad for this. Do remember if the power supply is remoted, even a couple feet, that these are 3.3v at 10 amp filaments, you'll need some really fat wire for that. The tube I happened to pull off the shelf is a samsung OM52 series for what it's worth -- a very used one. Since it worked fine, I've not run through a bunch of other types to test them for this, but since there's demand, I'll probably wind up making a few more for other people.

The quartz tubing enters the chamber through a viton o ring coupler I bought from Lesker, welded into a 1.33" CF flange (which was a trick indeed, but that's what I had on that end flange).
That end has a 1.5" long piece of Al pipe in it, most of that turned down to fit into the quartz, with about 1/4" extending from the end, so any sputtering that happens doesn't land on the tubing and make it conductive. Al is good for not sputtering BTW as things go.

At the other end of the quartz is a fitting I made of brass that holds an O ring for sealing, and is bored for a soldered in piece of SS capillary tubing that is 1.5" long and .007" ID, to control gas flow. This one's a little short and fat for that, but one I made on another source was 2" of .005" tubing and it was a little too restrictive, to give an idea of the range that works. You have to low temp solder this stuff as getting it hot enough to braze will cause some flakey oxide to form inside the tubing and jam it up, and it's a darn hard place to clean. The reason to have the gas flow orifice here is so that all the rest of the plumbing can run at or near atmospheric pressure, so all leaks are "out" rather than dirty shop air leaking "in". I have not noticed any sputtering off the end of the SS capillary tubing, but hey, it's at the positive polarity side of things, so shouldn't sputter anyway. Theory does work now and then.

Picture is a closeup of the homebrew hard line. As you can see, I had to make an adapter to make the 1/2" tubing fit into the pipe adapter, and slotted the latter and put a clamp on it to hold things together. I've never adjusted that length, it's fine just as it was put together.
Drilling that hole in the side of the main cavity was actually the hardest part of all this, not that turning the endcap of OFHC copper was a breeze, at least that part was thick and sturdy. Pipe is another issue altogether, and needed a good jig to hold it and careful milling machine work with very light cuts with a 1/2" end mill to get that off center oval hole made. I turned a piece of wood down to keep the vise from crushing the pipe while doing that. The end of the hard line that has to be curved to match the cavity curve was done in similar fashion on the mill, but really you could do that with a round file good enough without too much fuss.

Oh, there's nothing but air in there -- no dielectric at all other than 1.000000x air.
Attachments
HardlineDetail.jpg
HardlineDetail.jpg (62 KiB) Viewed 4180 times
Why guess when you can know? Measure!

lutzhoffman
Posts: 188
Joined: Tue Dec 08, 2009 1:59 am
Real name:

Re: Inverter Microwave for Ion Source RF

Post by lutzhoffman » Tue Dec 22, 2009 10:15 pm

Thank you so much for taking the time to explain everything, now I am begining to understand, at least enough to be willing to take a "wack at it".

Initialy I plan to run my accelerator at about 150-200KV, this is without any SF-6 insulating gas, with the tube in dehumidified air. The tube should handle this voltage in air at STP. At his low energy level I would like to be able to crank up the beam current big time, up to a maximum of 5-10ma for short times. For most experiments 100-500uA should be fine.

The next step will be to ramp up the KV to the 400-500KV range, this could be done with an SF-6 gas mixture at atmospheric - 2 bar pressure, now the max. beam current would be about 2-5ma pulsed even for experiments requiring high power. For only 2-bar, the pressure the vessel can even be PVC irrigation pipe, with an equipotential metal rings on the inside surface as well.

So in a nutshell the microwave ion source should work for me "as is" just like Doug designed it. I was thinking about a "hybrid" design which I call the CAD Ion Source (Carl and Doug). I am wondering if Carl's bottom end, and extractor, could be mated to Doug's microwave ion source. At this point I am not familiar with the bottom end of Doug's ion source so it is to early to really consider anything like this.

One thing I did wonder about is Vicor high silica glass? I know quartz is probably the best as Doug pointed out, but if Vicor comes close enough, and if it is much superior to Pyrex, then it may be worth consideration for some applications, since Vicor can be worked just like normal glass with oxy-propane. But then again it would probably be just as easy to buy a quartz test tube to begin with. Thanks......Lutz

User avatar
Doug Coulter
Posts: 1312
Joined: Sun May 27, 2007 7:18 pm
Real name: Doug Coulter
Location: Floyd, VA, USA
Contact:

Re: Inverter Microwave for Ion Source RF

Post by Doug Coulter » Tue Dec 22, 2009 10:24 pm

Here's the simpler of the two power supplies I've built to run this. I did use a nominal 4.3 kv transformer to run a nominal 4.1kv tube, but this may not matter too much. This could be done a little cheaper than I did; this was all parts I had on hand anyway.

I didn't modify the transformer, knock out the saturation inducing cores, or anything, just used it.
In the actual build, I added a wall wart and 3 terminal regulator for fan power uses. The Variac is adjusted to about 89 to 91 volts rms output, or about the place where the transformer itself begins to saturate with no load. You may be able to substitute a large incandescent lamp in series instead, but since I haven't tried that yet, I don't know what the wattage would be, probably on the order of 250 w.

You might be able to make the 2k resistor smaller (as in 1k), I made it big so I could light that big neon bulb current draw "analog meter" I used to monitor operation. In that huge neon bulb (got it at the last HEAS) you can see how much current by how much of the big electrode lights up, it's quick to glance at and verify that things are on the bogey.

In use, this results in somewhat lowered filament power and terminal voltage. That's needed to not burn up the tube which was never designed to run CW -- normally the pulse mode lets extra electrons get out of the gap area naturally and not bombardment-heat the filament too much, in CW they can build up and make troubles -- a reason no one uses magnetrons for CW work, or nearly no one. I have a 12 volt computer case fan strapped to the tube for cooling, and run another one across the general area to keep my HV feedthrough cool as it's designed to couple heat from the grid out of the tank (this will be another thread later). You *may* be able to run full line volts, but the transformer will then get hot, if you use a smaller than .1uf series cap here to limit tube current. But these MOTs run in saturation (as a voltage regulation feature in a normal oven) and get quite hot if run for long times without a fan, or really even with one. Here, we just don't need all that. In this design, nothing gets hot.

I also made PS on a ferrite core, using my H bridge design, but though it works, it's too complex for this, you just don't need it -- over engineering is as bad as not enough. And by the way, since those ferrites run at ~8v a turn, you have to then make a stepdown transformer and actually rectify the filament voltage, as otherwise the crummy lossy chokes inside the magnetron will burn up! Too much for something that is so simple, so I dropped that project. There's just no need for the complexity.

I will try to get cavity details posted soon, so post any questions you want answered, as that will be the last post I make on this -- I will include the ECR magnet design in that one. My next version will have the cavity inside the tank on a 2.75" CF flange/pipe so the ions can get in there a lot easier, as only RF and tuning need to go through the wall, both of which should be fairly easy (that is, if the type N weld in RF vacuum connectors I got from Lesker don't give arc troubles). In that design, there won't be the quartz tube, and the cavity bottom will be Cu or SS screen wire for the ions to come out, and the RF stay in. I am still pondering how I can do adjustable ion extraction field on that one with least fuss.

In my particular version of this, running up the input power makes it mode hop and go in and out of operation. This isn't a problem, as it has plenty of RF output well below that point. And of course there's that nice visual indication it's working. You have to see this with neon, it's blindingly bright.

I seriously doubt that you can modify an inverter type oven supply to do this -- they only go to half power on inverter PWM, then revert to on-off operation for lower powers -- I think the issue there is that the filament is run off the same ferrite transformer and it gets flaky at low filament voltages. That's 5-10 times the needed RF for this, and too much is BAD for this. You'd have to come up with some filament power some way, and fool all the fancy feedback loops in that inverter -- I've torn one down to see, then put it back together and sold the oven, it's just not worth it. They *are* great for cooking though.
Attachments
uWaveIonPS.gif
uWaveIonPS.gif (14.36 KiB) Viewed 4180 times
Why guess when you can know? Measure!

User avatar
Doug Coulter
Posts: 1312
Joined: Sun May 27, 2007 7:18 pm
Real name: Doug Coulter
Location: Floyd, VA, USA
Contact:

Re: Inverter Microwave for Ion Source RF

Post by Doug Coulter » Tue Dec 22, 2009 10:44 pm

Lutz,
you may have focus troubles unless your volt gradient is WAY steep trying to run that current. You need about 60x the gradient as for electrons for D ions for the same amount of beam blooming vs current. Depending, as always, what you call "focus". TV CRT tubes run a few ma at 30kv or so.....you can work it out from there. This is because the D ions are going slower, so for a given current and volts, there's more in flight next to one another than for electrons at the same volts and amps, and they have more time to repel one another on the way to the target.

Glass "alloys" are just about as diverse as for steel, it's astonishing how many there are, all characterized in the Kohl book on "materials and techniques for vacuum devices". Or maybe even not all, but it's a heck of a loooonng chapter.

Vycor is one of "those", there are "flavors". The top grade is quartz, plain and simple -- no difference.
Some is kind of porous, and I'd not use that for this, but you have to ask the source supplier.

The reason for this is quartz is far lower loss than pyrex glass (DC and AC both), and more resistant to being taken apart chemically by hot hydrogen ions. Alumina isn't even in the running for this, too lossy by far.

Here is where we get ours, a small outfit, and nice to deal with too -- they like big orders, though, so people may want to get together and do a group purchase (I will do one for this area if people want me to -- I am using the stuff up and want more myself).

http://www.quartz.com/

I also use their stuff to make my own HV feedthroughs, having gotten disgusted with the commercial versions.

I work this with either a small oxy propane torch I made using a wire welder tip as a gas tip (works great!) or an oxy-acetylene welder for bigger stuff. You need danger-sensing sunglasses as it has to be white hot indeed and is blinding to work on. The regular "didyium" glass blower's glasses aren't the thing here, they only block the Na lines and you're still blinded. Welding goggles are about right.

This design needs *none* of that, I guess I wasn't clear enough on that yet -- more pix to come. It's just a cut piece of tubing, easy to cut another should replacement be needed, on the lathe with a tool post grinder and diamond wheel. Quartz you can't just score and break like glass. It comes apart at random if you try, and you might get cut while that expensive stuff gets ruined. I don't even flame polish the ends for this.
Why guess when you can know? Measure!

User avatar
Doug Coulter
Posts: 1312
Joined: Sun May 27, 2007 7:18 pm
Real name: Doug Coulter
Location: Floyd, VA, USA
Contact:

Re: Inverter Microwave for Ion Source RF

Post by Doug Coulter » Wed Dec 23, 2009 12:09 am

OK, here are the rest of the details needed to replicate this, on a virtual silver platter.

The cavity is a piece of 1" ID copper pipe. It is 2" long, but the effective length is reduced a little by the tuning cap depth of .2365" sticking inside. The tuning stub is a brass screw, 1/4-20 threaded into this to stick in .9860" for resonance in my setup -- that should get you within one turn if you're using quartz tubing for the gas.
The top cap is held in by tight fit and by 3 4-40 screws threaded into the pipe. There is a spring under the tuning screw head to get rid of backlash and so on. You will note the screw has it's threads taken off the end, and the end rounded to cut down arcing issues.

The bottom is a pipe cap, shortened and held on by tight fit and 3 4-40 screws threaded into it.

The gas inlet is a piece of 1/16 od, .007" id SS cap tubing, in a brass fitting I turned down to fit, which seals to the quartz by a viton o ring. The other O ring you see in the picture is just a cushion so I don't stress the quartz (it's rough end, I did not flame polish it or even grind it flat) -- this makes it trivially easy to replace if needed, but this one has run over 100 hours and is still pristine. This is charged to a positive voltage for ion extraction, via a CCFL inverter that has variable DC input to control the extraction voltage. This supply is bi-polar (two half wave volt doublers off the same source, opposite polarity), eg if it's making 1kv positive, it also makes 1kv negative for the other electrode. I use about a 20k 2w resistor for protecting this from arcs from the nearby HV tank input.

Not shown in these pictures is a polycarbonate 1/4" thick shield I also use to keep arcs away from this.

The magnets are NdFeB from Amazing Magnets, with a half inch center hole. I used 1/2" by 1" to stick them to, drilled for a little over half inch to let the tubing pass, and 1/8" by 1" iron for the connecting yoke. You can see in the picture the shim I taped in there after measuring the field with a hall effect DC magnetometer to 980 gauss or thereabouts. The field isn't very uniform, but seems fine, and really makes a *huge* difference in how well this works at very low pressures. Once lit off, I've seen it still run down to 1.4 e-6 mbar, which happened to be the base pressure of the system at that time, due to air infiltration through the necessary insulating silicone tubing on the gas input. I may try another flavor for that piece, but it has to be there to run the gas inlet at high positive voltage to repel ions into the tank proper. With ion extraction volts at max, about 2.5 kv + and -, and a pressure anywhere above e-3 millibar, this lights off everytime just by turning it on. No adjustments needed.

Not shown, as the tank is under vacuum just now, is the other end of the tubing. At that end, there is a piece of 1/2' OD aluminum pipe that is turned down to fit inside, all but about 1/4" of it, and the end is about flush with the magnet closer to the tank. The other end projects about 1/4" into the tank past the end of the quartz tubing, so if it sputters, it won't sputter onto the quartz. This is hooked to the negative polarity of the ion extraction supply through the same value protection ressistor, with a wire run inside the tank and insulated with alumina beads -- you can see it in other pictures I've posted here on fusor operation, glowing red in X rays.

The hole you see in the cavity in the last picture is for tuning the coupling rod length. It's useless more or less, even a teflon tweaker mis tunes this so bad it's not funny, and burns up in the bargain in the field. So that part was tuned with power off and try a little this way and that until it seemed fine.

So, grab that paper linked above, and this thread, and doing this will get you a very versatile and reliable ion source of mostly monatomic ions for your fusor or accelerator. As I said above, the next step after this is for the cavity to be in the tank itself so we maybe don't even need any explicit ion extraction field at all -- the grid field will probably do fine at that point.

So, that's it, any questions?
Attachments
Magnets.jpg
Magnets.jpg (46.03 KiB) Viewed 4180 times
GasDetail.jpg
GasDetail.jpg (46.87 KiB) Viewed 4180 times
TuningStub.jpg
TuningStub.jpg (45.72 KiB) Viewed 4180 times
Why guess when you can know? Measure!

lutzhoffman
Posts: 188
Joined: Tue Dec 08, 2009 1:59 am
Real name:

Re: Inverter Microwave for Ion Source RF

Post by lutzhoffman » Wed Dec 23, 2009 9:36 pm

Hello:

Thank you so much for all your info on the ion source, I think I have enough to get going on it when the time comes. I checked with NEC reference the ion current which the tube can handle, and they put the max at around 25-50ma with the main limitation being the ion source flooding the vacuum. Thank you so much for your help : ) I am curious about one remaining thing which is the extraction end of the microwave ion source? Is it the same as Carl's, or do you just have a small hole at the other end of the quartz tube, with an intermediate extraction electrode close by at - potential? If I understand correctly this is where the CFL inverter comes in?

An interesting feature of the tube is built in intermediate focusing electrodes which are located between each accelerating section. Basicaly they are the same as the first elecrostatic focusing lens which accepts the ion beam entering the tube. This gives some control of the ion beam between sections. The entire assembly contains 28 electrodes including the entrance, and the other 2 focusing electrodes. My plan is to drive the tube with a 14 stage FWCW multiplier, with 13 equipotential rings, the terminal being connected to the 14th stage . I am considering two options on how to connect the tube to the VM, the first is:

Every other electrode would be connected to the 14 stage VM directly , via a HV resistor. Then the electrodes in between these would be connected to the driven electrode below it, also via a different value HV resistor. The question here was do I connect the 14 intermediate electrodes to the VM connected electrode; Above, or Below it?

The second method being considered is to simply build a 28 stage resistive divider, connected at 28 points to the tube, and at 14 points to the VM stack?

The first method would seem to be more efficient, but the second would seem like it would have a better potential distibution. My plan is to build the first 6 stages, and then run the stack at a reduced voltage, and do a bunch of testing : )

The actual VM stack is fairly well protected with series resistors (About 5-10K ohm combined value) between each stage capacitor, in each stack. This is in addition to all of the normal protective measures like spark gaps, input resistor ect. I am also using 3-4 times the nomal number of HV diodes, in the first, and in last stages of the VM (In parallel for higher surge current rating) since these diodes will see the full surge current if a discharge were to occur, which would only be limited by the series resistance of all of the 5-10K ohm protective resistors.

Thanks.....Lutz

Post Reply