I've been working on a new ion gun design which seems to be working quite well, however it has led me to discover something very interesting that's going on, and I'm wondering why it's doing it.
The ion gun design is a KF50 flange with about 150 turns of magnet wire around the tube. At the time of this incident, it had about 6 amps through it, so the field at the center was somewhere around 200 gauss (very much an estimate since the exact winding count is unknown and there was no active current meter on the DC supply at the time, but after previous probing and knowing the variac setting, it was about this amperage). Then there is a little mini-grid inside the flange attached to a feedthrough. This then connects to a rectified NST. So, basically a pretty simple DC magnetron using an electromagnet to vary the field.
So, what is the incident? The first observation during operation around 10 microns D2 is that there is crackling noise coming from the chamber right when plasma forms at about a kilovolt positive polarity on the feedthrough. This makes massive RF noise, enough to throw my neutron counter to ridiculous counts and even turned my computer off on two accounts. Interestingly, this interference and crackling noise entirely stop as the NST variac continues to be turned up, then as the current/voltage diminsh coming back down it again starts up until the variac is turned to zero. But the incredible thing occurred when the wire providing the feedthrough voltage arced 1.4 inches to a ground sleeve. This thing was being fed by a center tapped 12kV NST with the variac at 25V, so no more than about 1.5kVDC there. I can't figure out what's going on, there seems to be some sort of charge establishing past the rectifier, but it just doesn't make sense to me. This kind of an arc would take 20kV+, not even remotely close to the arc ability of 1.5kV.
So the question is, why is there a massive charge building up here? My only thought is that charge is being pulled into the grid similar to how a van de graff picks up charge without giving it back out the source, but since it's all electrically connected here I'm not seeing how that's possible.
I've attached some images of this, the first being the electromagnet, second showing the arc, and the third showing a diagram of the wiring and coil.
Well, thats somewhat a new one on me, but I have seen some very interesting things here, and blew a NST via internal arcing on the last run doing something a little similar which will get it's own post/topic. Or this could be the start of something bigger that winds up in fusor theory once we get a better clue.
I have noticed two interesting things that may be relevant, who knows? One is that using a basically diatomic gas, you can get some very weird things to happen. At some point, it's as though you had made a large polymer of gas atoms, that wants to revert to "G"2, and if it's that, try nitrogen -- it's far more dramatic. This makes noise even in a pretty good vacuum, a bright flash (inside the tank) and so much EMP that I've had proximate computers not only crash, but permanently wrecked. When I mentioned it on the fusor forum, I got a lotta BS about it, and talk about charge building up on isulators, but it wasn't that (orders of magnitude too many joules) -- the flash was in the bulk gas in the tank, and blindingly bright. I will look for the movies I've taken of it. The only other report I've heard of this which says about the same thing, they were at the other extreme conditions -- hot N at mega-bars in a diamond anvil cell. I was at cold N (Joule-Thompson cooling from a jet, ionized on the way into the tank) and low pressure and just running a 10w input ion source, nothing else. About every 10 seconds, I got this bang of about 100 joules -- that's not the kind of energy that will build up on a sheet of glass! Sounded like the tank had been hit with a ball peen hammer, and blinded me.
The only time I've seen volts go *up* from the source outside the tank was under different conditions, though, and with D. I was driving a second grid in the tank with an NST through a small .01u cap, and using the tank itself as a diode to ground. This worked well as a half wave volt doubler -- it would clip at gnd trying to go positive as the second grid sucked electrons out of the plasma, then go negative. During the negative half cycle, I'd see pulses on the FT terminal of +20-30kv, while the NST was trying to make about 20kv negative! This of course resulted in the death of that NST as the resulting 30-40kv across the internal ballast was a bit more than half a winding could stand off. But, for what it's worth, that second grid with minimal watts on it pulsing like that -- drove the neutron counters, and more importantly the silver activation wild, huge gain in Q for the main grid output (or both, who knows, not I at this point).
This video shows what I was doing at the time, not the same as you, but a similar thing going on.
Scope trace - neutron counter upper trace, voltage on secondary grid, lower trace. Lower trace 10kv/div.
Here, the upper trace is the output of the 3He tube (after some pulse stretching, a transformer, and an audio amplifier, that's why the weird jump up after the first negative pulse, which is energy stored in the isolation transformer and the audio preamp clipping -- for what I'm doing, I don't care about that).
The lower trace is the realtime volts on the FT using a frequency compensated high voltage probe. Zero volts is one major div below the middle. There are several important things to notice here, I think. This is chopped traces triggered off the 60 Hz line, so within the camera exposure time -- they are concurrent. Two things that are making me really scratch my head here. One is that you can see that when the phase is such that we are trying to drive the secondary grid positive, it takes awhile for it to start "rectifying" and clamp the volts back to ground -- eons in electron transit times. Why? Second -- why does it take so long after getting to negative volts for those pulses to begin? Eons again, and so is the repetition rate, slow considering the equations of motion I think are applicable. And why do they go positive of ground?
Now, to add something else I don't right now have a picture of (didn't take one before frying the NST, Bill is bringing another this week) is that the faraday probe shows the plasma net negative a couple ms (during that positive overshoot on the second grid) and then going positive as the 2nd grid stays in the positive going phase, but is getting clamped back to gnd -- for a couple ms. After that, the signal on the probe goes positive about as far as it had been negative (about a couple hundred volts from a wire in there into a scope input at 100 megs load). It doesn't seem to see those sharp pulses that the other things see, though, during the 2nd grid negative half cycle, and I'm not sure why other than it's about 4" from the action.
Without that second grid connected, I always see a couple hundred volts net negative on the faraday probe.
This is puzzling me no end. The only thing I can think that makes sense with laws of physics is induced charge as a bunch of pure positive ions come into proximity to the grid on the way through -- after all, the neutrons seem to be coming out in bursts right then. All the times involved, though, are EONs compared to what you'd predict from the speeds of the particles at the assumed voltage fields. Milliseconds, not nanoseconds. But why in heck does it take so long to happen? Something else (perhaps some odd recirculation) has to be going on first that I'm not seeing, first? Something that eventually builds up to some sort of avalanche condition?
Note, you'd not necessarily see this if I ran my 3He tube normally. I have a very high threshold set so I only see a pulse if it gets hit by multiple neutrons inside on tube response time -- more than one per 10-30 us in other words. I do this as I'm interested in bursting, and because otherwise it always sounds about the same on the audio amp you can hear in the movie it counts so fast.
My pulse widths have sub microsecond rise times, and about 10 us fall times, but only occur at sub kHz speeds. It's like it takes a long time to "make up it's mind", then avalanches, and does this a little *after* you get to negative volts on the thing, after having sucked out the electrons and made the plasma net positive (we normally measure net negative with a faraday probe in usual DC fusor operation). I'm getting about 5-8 pulses per 60 Hz half cycle doing that, at least until it fries the NST.
At any rate, I'm seeing something like this too, and it seems like it's worth serious investigation -- because of that neutron output -- I could easily believe that by some magic this level of EMI is getting into a solidly sealed metal, battery powered internally, preamp box on my 3He tube, except -- that silver got hotter than ever, it was a record run, and you can't cheat that one! During this, my little ion source was running smooth the entire time, making about 2ma worth of ions into the tank. Pressure on my gage around 1e-2 mbar, I was varying it from half to twice that at times. (gage reads high on D and higher yet on already ionized D).
Having said all this, there is a potentially simpler explanation for what you saw. There is so much inductance at the output of an NST that you might just be having plain old kickback if the discharge is unstable. In a car spark coil, a 12v input on the primary can kick back to 600 volts when the points "open" -- and an NST is a much better inductor. So you'd have to put a scope probe nearby (not too close!) and look and see I'd think to be sure of what's going on there. One way to test that might be to add a fairly substantial filter cap, then a series R to the ion source and see if it still happens at any point in the circuit.
While you can't count on any particular frequency or time response other than it will be differentiated, a 10x scope probe a safe distance away (say 6" or a little more) will see a lot of this and might give you a clue. The signals from this will be a lot louder than the usual room hum/local radio station pickup. Try it as a poor man's probe for things like this.
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.
After my original post I did some more work on this ion source and came to an interesting observation when I started filtering the magnetic field. The original issue persisted at fusion pressures no matter what I did, so this was basically useless since all the arcing caused instrumentation failures. When I added 100,000uF filtering to the magnetic field, the issue completely ceased and the ion gun operated as expected with no high voltage build up or arcing issues.
I'm not sure exactly why the 120hz (rectified and non-filtered) electric field would cause this charge buildup to occur, it's actually quite an interesting phenomenon in itself. But without the pulsating field it can't build up.
I'm going to look into what higher than 120hz (potentially into the several mhz range) forced polarity reversal on the field does in December (off at MIT for now so no more work on this project for the time being).
Guys once you start an ion gun, charge is transferred and various things will charge up until they breakdown --and this to way past what acceleration voltage you are using. The next thing is that some gasses do not do negative ions at all well so if you have alot of AC on your acceleration voltage funny things happen as your ion flow is unidirectional ie a diode of sorts.
Ps this happens on a good DC supply with alot of arcing inside your system and things you thought were not in the way of the ions charge up until breakdown Ions have a lot more mass than electrons and travel fairly slowly at the voltages we use so what happens if you try to change its travel direction half way though its mission ---it gets confused --stops--gets bumped by other confused ions and quite few end up in a totally new direction
From memory Nitrogen and helium are buggers for doing this if present
In your cases there should be no helium but there will be enough nitrogen
I would think that once a surface charge built up on an insulator that Coulomb's laws would still work and repel further ions, else we've got some sort of perpetual motion machine? At any rate, I've never seen volts on a faraday probe exceeding the acceleration voltages in use, nor had breakdowns from that. I have had them from hot ions chemically reducing silica to silicon....and some exceedingly weird energy storage when testing sources with N2 (less with H2 but still there). That latter was evidenced by a very bright flash and EMP in the 100 joule range with nothing but an ion source with 10w DC input having run in the tank for about a minute. I think that was some sort of chemical effect of all the N+ getting back together into N2 or something similar.
I mean, if I've made ions go say 4kv (the actual case in my ion source) and I put an electrode out there at +4kv, it stops them from hitting it -- they might go around and go elsewhere, but I stop seeing any further current when I get up to the same energy as the beam. Below 4kv I start seeing positive ion current on the probe, above, none.
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.
