Microwave Ion source

[Doug Coulter]

They are overkill, and I had them, you guessed right. Regular microwave diodes are fine (but you might want to use two in series for the second one in the circuit). I'd bought a couple hundred of the ones I have as they're nice for HF HV supplies in the multiplier part. 1n4007 diodes (seriesed up) are a little on the wimpy side for the startup surge, I had these, it was done!

[JonathanH13]

Latest power supply schematic for ion gun.

I have made some small changes here: I redrew the transformer, showing the internal ground connection to the transformer core, on the output side, not the input side, which was incorrect. I also removed two 10 megaohm resistors, to drain that large capacitor faster. I found that a 1uF cap @ 9KV will suffice for C2. I also reduced C1 to around 0.047uF @9KV. So, to be clear, the top circuit is the latest one.

[Doug Coulter]

Thanks for the neater and more readable version, Jon -- this is exactly what I have going here. You will also need a positive polarity supply for the ion extraction electrode (pusher in your case) which I built off a CCFL inverter/multiplier so it's separately adjustable -- you don't need too much current for that. But thinking about it - once you get the voltage right (or know what right is) you could probably do that off the original transformer as well, saving some parts and money. Diodes that go 50-60hz are a lot cheaper than the fast ones needed for a CCFL after all. For that, a bunch of dirt cheap 1n4007's in series would do just fine and the caps need not be so large.

Since 5w neons are quite rare, you could use a smaller one there with a larger resistor in series with it. I used that big one because I had it, and because it's a kind of dirt cheap milliamp meter I can read from across the room -- the lit area of that big plate is more or less proportional to the current being drawn through the big filter resistor. If you have a real milliamp meter, a 50 ma full scale is about right there instead of the big series R -- you'll run below that, usually 30 ma or thereabouts. But in that case, you'd have to devise some meter protection as the transformer will easily put out an amp if the output is shorted...where the neon just lights up bright.

I'd note that the bleeder shown is what I have -- but it's really inadequate to discharge that big filter cap quick enough for safety. I just didn't have power resistors of the right value so I used a bunch of little ones that don't really draw enough current. Anyone building this should make up a chain that can draw a couple milliamps -- about 10 watts at the full voltage. The charge on that filter cap is wicked. I'm lucky to be alive after getting across it a couple times before I put it where I can't get to it easily anymore. So instead of 40 megs with 1/4 watt resistors, you should have more like 2 megs with say 10 watts dissipation capability (at least). In my case that would have spoiled the purity of being able to make the thing entirely from parts on hand....duh.
Duh, because you feel really stupid getting the daylights shocked out of you minutes after power off

[JonathanH13]

It works!

I have built the power supply for the ion gun - it worked right first time (that always makes me suspicious). It also seems that the ion gun survived being rough handled in my luggage between US and UK

I connected it up to the vacuum chamber, and pumped down to around 1e-2 millibar. I then connected the fusor power supply to the terminal at the rear of the quartz tube, and put about -4KV on there. This created a nice plasma in the tube. I then turned up the voltage on the magnetron variac to around 180VAC, which is around -3.6KV on the magnetron. It suddenly burst into life, getting brighter as I hit the 'sweet spot' and then started to mode hop as I went above that.



I checked the new ion gun for microwave leakage - it leaks a fair bit (more than my leak detector can count, when held close). That is over 10mw/cm^2. The warning alarm triggers at 5mW/cm^2



This system also generates a heap of UV, which is always nice - 5 minutes in front of this feels like 2 hours in the sunshine. I think maybe it's time to put up some screens...

[Joe Jarski]

Nice work! And thanks for the updates on the power supply. I think Doug had mentioned going away from having the gas flowing through the gun - is that your setup?... kinda looks that way. One of the these days I'll get mine going, I just need to get the string of diodes and put it together. Maybe I should invest in a leak detector too... Sheesh, slow down self!

[JonathanH13]

Thanks Joe. Well done with your ion gun - it's a thing a beauty! I was just recently admiring the precision machining you did on it.

Having the gas flowing through the gun means floating the entire gas system at high KV, unless you are prepared to make a non-conductive coupling that does not leak deuterium.

At the pressures we run at, in the molecular flow regime, the mean free path is much longer than the dimensions of chamber. So the gas molecules are acting independently of each other - they randomly make their way into the gun (or are attracted electrostatically), get ionized, and then are repelled by the positive potential. In Doug's original system he had a +2KV push on the terminal side and a -2KV pull on the chamber side. I'm not sure yet if I need this, or if it will work OK with a +4KV push on the terminal side and a -40KV pull on the grid. I think flowing the gas through the gun produces more ions, but so far it looks like we can get away with this, which is a lot simpler. Doug and I also spent some time discussing a 'differential' flow design, which sounds awesome, but would add further complexity.

[Doug Coulter]

Glad to see it made the trip -- these should, they're pretty solid the way we made that one.

We did go away from flow-through gas. I'd made a coupling of pyrex tubing a little over a foot long to isolate the gas electrode when I did flow through gas, but under some conditions that would light up and short the push power to ground, so I quite doing that. Since I've gone to batch mode, the pressure in the supply line is about atmosphere during a gas pulse, but then goes down to eventually the tank pressure as the gas goes through the thing. At some point it finds a Paschen minimum and lights up like a neon tube, shorting the pusher supply to the grounded tank plumbing. Never seems to happen at a convenient time. In experiments, I found out that once lit, you didn't need the extra gas in the source to keep it lit if you were willing to fire it up while the gas pressure was still on the high side (e-2 or e-3) and it would still stay lit to almost e-6 mbar soooo...simplification is good!

You do get some short UV off D, but it should not be much as you should only be making ~20w of RF in the first place and that just won't make that much light. A regular-glass or plexiglass shield should eliminate concerns.

I don't have a lot of microwave leakage here on my (cheaper) leak detector. I'd guess either an endcap is a bit loose or that the length of rod for the pusher is resonant at this frequency if you're seeing a lot. On the other hand, mw/sq cm isn't a lot if it's in a small place and you're not pressing your eye up against it. Were it a watt, I'd be more worried. It obviously matters a lot more if it's high where you are at, or jams your wifi (same frequency, which is why it's used for that - no license needed).

One thing most people don't realize is just how low the percentage of ions is even in this, or in a fusor running full bore. As an experiment sometime, let in just a little too much gas (on my gage that would be about 2.5e-2 mbar). This will cause any reasonable current limit to be hit, in my case around 20 ma at about 40kv. Well, I have this super high power supply so I can crank up the current. Guess what -- the voltage won't rise even going all the way up to 50ma - you just ionize more of the neutrals and the voltage stays depressed!

That's one of many reasons I'm going to build a deliberate beam-linac device, so as to run in "pure" ions, not a mix of ions and neutrals and electrons.

[Joe Jarski]

I'm almost done with my uWave power supply. Everything is working good so far - I get about -5kVDC out of the filter cap on the top end (120VAC input), which falls in line with the others - plus I have a smallish magnetron. Although, now I have a question before I finish things up. Do the "F" and "FA" connections matter in this case? I'm thinking it doesn't because we're running these CW so the phasing shouldn't matter like it would in a typical oven running in pulsed mode, but I just want to check. If I need it, I've got it written down somewhere... could take forever to find though. He's a picture of the guts for everyone's amusement...

[Doug Coulter]

Filament phasing hasn't been an issue in mine, I didn't bother to even check. Some MOT's have two secondary taps (for higher or lower line volts) and that would matter some. We're running our variac at about 89-91v output, just at the point where we have enough filament power to make it go. It's pretty stable there, it's right about where the transformer core begins to saturate (which also keeps the transformer from getting too hot in full saturation). One could probably substitute the variac with an incendescent bulb, actually, but I've not worked out what the wattage rating would be to give the right output. I plan to try that at some point, as the load is friendly to that approach.

[Doug Coulter]

FWIW, I happened across this video while looking up the magnetron in an oven I just got from the dumpster.
http://www.youtube.com/watch?v=m6aZaIqy2rc


This shows a magnetron on a spectrum analyzer. Look at that sucker mode-hop.