Science/Engineering/Technical forums

And here we have some measurements of that. I showed the basic setup in eye candy. I then let in some gas (no change just doing that) and then turned on the ion grid. "Jaggies" appeared on the swept impedance plot, which in the case of the fist measurement, slowly went away with time (maybe half a minute or a minute) while the apparent resonance moved up somewhat from what it was at first. So, I tried both more and less gas with ions on. The upshot is that more gas means it looks like a larger capacity, and for the most, those jaggies didn't go away. (I was using fairly decent DC on the ion grid from a Spellman 250w supply). The VNA signal isn't enough to keep things lit on its own, and in fact this is why I got that Analog Discovery 2 from Digilent - I plan to use it to drive a power amplifier and get the same (more or less) measurements "at power", which will also be interesting in terms of "when do ions/electrons start hitting the tank walls" kinds of questions.

I got faked out by the marker (listed as #1 onscreen) moving because a trackpad accidental click instead of the drag I was going - I hate trackpads (and they do clicks even with that function turned off often as not), and in this case, the darned unreliable human observer didn't notice the reference mark moving and took wrong notes in the screenshot!
So pay attention to what the VNA plot is saying, not my little window with text so much. I was wrong - it's pretty much a case of more ions, more capacity (or at least lower resonant F). Now I added some gas and turned on the ion grid with DC, a couple mA on that, voltage depending on amount of gas (current limit situation). Incidentally, I noticed a net negative on the faraday probe, which is normal for DC ion generation. I should try with RF (well, 45kHz) ions, but that would probably drive the VNA completely crazy. And so on, more and less gas, and while I did write that correctly in the gedit window in the lower right, I failed to notice my reference mark for resonance/no gas moving on me, so I told a lie there. Which I note that despite my incorrect note taking that where we usually run with pure DC drive - doesn't affect the AC impedance much at all....
Which could lead to all kinds of speculations...I think looking at the Q (in the RF jargon sense) is important, and this looks to me a little like how a mass spectrometer goes south with higher pressures - only here we only start to see that at ~~ 10 times or more the "pressure".
It's that effective capacity I'm interested in here mainly - it means I can transfer energy to and from the ions, both....scattering should show up as plain old loss, right? But this means if I can set up a recycling situation, I can get a lot of the input energy back from the "flywheel" that is ion motion in and out from the center....we'll see, early days yet.

Edit: note impedance in the last one is around 2x what it was at resonance in the first one, even though the resonance wasn't much different. Losses due to ions/electrons...now we can start to
do some interpretation.

Which could lead to all kinds of speculations..

It did, but the thread title seems obvious now. A plasma under an electric field should act like a capacitor. At least in one direction. Right?

I don't know enough yet to make good speculations.
For example, I can measure a time delay between driving the main grid and another probe about 6" out pretty easily, a delay that is there when ions or plasma is there, but no delay (capacitive coupling) otherwise. (speaking of transit time here, not a load on the main grid that "looks like" a few pF capacitor)

Now - when the main grid goes positive, and a little later my probe does, is it because:
1. main grid repelled positive ions and that's the flight time to there
2. main grid sucked up electrons (we see current draw) and therefore the whole plasma became more positive after n electrons were sucked up.
3. main grid made a field that polarized the plasma - all the + stuff of any deuteron/electron pair went one way, while the negative side went the other and it took time for the rotation to happen.

Or, all of the above - still sorting that all out. I know what I think, but in this game it's what you prove that counts.

Important question is "why does this work at all, when:
As John Futter once worked out, an actual serious charge separation of that many charges would take the energy of a hand grenade - we're likely not dealing with just + or - particles here. More like a soup of some sort. He's working with microamp "pure" beams - those are controllable, but he can tell you how hard that is; we're in the milliamps here.

The pressure is ~ 100x higher than where mass spectrometers utterly crap out due to collisions and random scattering.

This is all a super sensitive function of "mean free path" it would seem. In mid-range vacuums, Lesker and other guys publishing math on behavior just go nuts in the area we inhabit. In really good vacuum - long mean free path - it's all billiard balls and not hard to compute. In really crappy vacuum you can talk about pressure, sound, hydrodynamics and stuff like that - and it's easy to compute,mostly (but ask the tokomak boys about oscillations and waves and junk in system that creates forces that react back on the system...). OK, so now we add charges just for fun, and those see each other all the way across the tank regardless of MFP - to a point (there's this thing called screening for some cases). And then we work in an area that confuses heck out of guys working with a much simpler system with no "extra" forces like Coulomb happening.

Now what I'd like to believe...
From some other measurements it does look like indeed we can have fairly good charge separation at fairly reasonable pressures (here, say 2e-2 mbars indicated) - more than you'd think, no hand grenades involved. This does seem to be about the limit of "minimal smearing out due to stuff banging on other stuff". But I was still seeing reasonable behavior at a couple times that...I need more data! (always, what else is new?)

It's long-known that particles going past an electrode in a vacuum induce charge on it as they arrive and then as they leave (opposite signs) - and that's what I'd like to believe is going on here.
I should pull up the relevant link from Humphries' papers on accelerators in our /data section to back that up, but it's well known in the biz.

What is NOT well known, is that we can do this "at pressure" and believe me, compared to where the mass specs and accelerators and so on operate, we're "at pressure" - and if other measurements made here are right - we have a "compression ratio" and vastly different pressures at the focus than at the edges, even with DC drive. With AC/dynamics it should be possible to get a far higher ratio...

So I imagine this:
We have a more or less random distribution of charge. With this ion generator, there are a lot of extra electrons, but other than that, in other tests, I see about the same stuff with an AC ion source, just less negative bias on all the probes.

When we drive the main grid negative (for example) - the + charges start moving toward us. At first current drain is high as we're putting on a field and they aren't moving yet. As they come up to speed, the grid draws less current (just like a capacitor would as it charges). The same thing is happening in the opposite direction with the electrons, only around 60x faster (it's the sqrt of the mass/charge ratio - ratio, which would be 1837 or so if our + was a proton but is twice that if a deuteron more or less. So sqrt of 2x1837 = 60.61 roughly). For this part of speculation, let's assume the electrons just hit the walls, mainly.

As the + charges move toward and go inside the grid, it ceases drawing current from whatever was driving it positive. As they leave, the opposite occurs (Humphries). When our main grid goes positive, it's just like the plate in a vacuum tube diode - electrons rush in and being fast, even if they have to spiral around a few times, they hit the wires and draw current real good - we see this effect when we capacitively couple an AC signal into the grid - it acts like a diode to ground, a pretty decent one - in DC drive, the grid gets hot and releases secondary electrons as it gets hit and we can see that some materials release more (just like was worked out in electron tube design). So, in general, while the ion grid is pumping in extra electrons (a power supply is after all an electron pump) we are eating them, and one of the measurements I should have taken today would prove that - I'll get to it next time, but this action should "eat" electrons from the excess created by the ion source (and I'd bet it does, didn't bother to look this time is all).

Basically, the ions are seeming to act kinda like a mass in the spring-mass system, and as long as we don't drive them into the walls, or hook them back up with electrons and make them neutral again, we can "juggle them". Due to geometry, there's a lot more density or "pressure" in the grid focus range than elsewhere, but the pressure there and outside the grid varies with the RF drive phase - minus some lag as these things have mass and respond accordingly. The spring of course is our applied (or more likely, net) field...and it seems we can *almost* ignore the ion fields up to around ~ 2e-2 mbar, after that it starts to make a bigger difference, and by twice that all the dynamics get smeared out. Which makes some sense, or at least it does to me.

Scattering of ions off one another would look like a loss in this system, as the regular back and forthing would be somewhat suppressed - there would be time-smear as some went faster and some slower. So it looks like a resistor on our tuned circuit...OK, fine. The "intertia" has some drag, or friction. Works for me to understand it.

What's potentially very interesting is our geometry here (let's pretend we have no end effects for the moment). So, ions head for the center, some scatter but the net energy of them coming out, whether scattered or not - is the same! So what if some come in from the north and exit from the east? If some go faster, some other one goes slower, so what? Same energy net! Due to circular symmetry, if we can get them to linger outside the grid near the tank walls (where they'll be going slower anyway....hmmm, yay!) they'll re-uniform their distribution by repulsion and be ready to make the next cycle identical to the last one - all our scattering "losses" - the bane of this kind of fusion - are eaten up at very low per-particle energies, not wasted at high kilovolts!

So, what a lot of this has been about is figuring out how to make an ion trap - much the same as what goes on in a mass spec, but at higher density, but instead of rejecting ions that aren't just so, we
want them all contained, and hauling enough freight when they get to the focus to fuse - so we have to figure out a speed/feed that gets them hauling ass but stopping before they hit the tank walls (which is where most of the loss is in a normal fusor). Juggling. At times, we want all of them, or as many as we can, colliding at the focus. Rest of the time, they're slowing down as they head for the tank walls, giving their energy back to us, and getting ready for the next roundy round. Eg, there is no one pressure at any one place...it's all dynamic with time and place in the tank.

Now, what happens in a "normal" ion trap is that after awhile, the few ions that remain have "learned" or better "been selected" to have trajectories that are the lowest system energy, which is a fancy way of saying "they don't collide" - but that's after tens of milliseconds in a multi megacycle system...we want our fusion to have happened long before that many cycles! It's just that selection process that bumps them off one another that is our desired effect, instead of something those ion trap boys try to avoid...and we don't have to take care about rejecting ions with a slightly different mass/charge ratio as we're not measuring that at all - we have 1, 2, 3 and stuff like that - 2 is a heck of a lot different ratio-metrically than 1 or 3! We're not trying to measure the difference between a deuteron and two protons, or one unbound proton + one unbound neutron - we don't care.

The really super good news that no one yet has noticed is that even at a mean pressure 100 times higher than a mass spec will work - this works fine. Even to 200x, then it gets crappy. But! We do have a compression ratio here and it's huge and that's not counted in the 100x....heh. I'm staing inside the "region of stability" instead of borderline - see above, I don't want to reject anything that might fuse!

I do have another trick or two up my sleeve, regadring electron tricks at just the right moment as the "spark plug" in our "cyclic engine" that's starting to almost look like it was inspired by Otto...but that's for another day.

The only think I have to back any of this up is intuition, and the test that showed us real fusion happening with AC drive at *far lower voltages* than ever before measured here. But only with AC, same everything else.

But it's a good place to be - it passes the Feynman tests, it explains observed stuff and also gives me more stuff to test and more predictions.
It's worth that cup of coffee. Or really hot tea.

Nice to see some data! At a glance, this looks to me like Landau damping, i.e. wave damping on the electrons without collisions or dissipation. The waves are giving their energy to the particles, just like you would want, except that it's probably the electrons getting boosted rather than the ions due to their low mass. The damping should increase with density, as you're seeing, and is important when the plasma waves have a wavelength on the order of the Debye length or shorter, as you very likely have. The thing that doesn't seem to make sense to me is that your resonant frequency appears to be going down as you increase the density - am I reading that right? For plasma waves, all else being equal, you should get the opposite.

Nice call, Paul. I looked it up, but tte field equations made my head hurt.

I guess you and Doug can do Navier-Stokes analyses in your head, but not me!

Paul, I don't think so...But that's why we do the lab work. Remember, mean free path is right around the size of the box here - funny things happen - the interactions are weak, not like a fluid with the concept of "pressure" at all - grey area between ballistic and viscous, more on the ballistic side. Of course, mean free path is really defined for uncharged stuff. We've yet to have that - or even a neutral plasma, we usually have a huge excess of electrons unless I put RF on the grid, and only that grid, then I see closer to neutral as the positive peaks on the grid draw current and take the electrons away. This is a real pronounced effect.

This one's kinda hard to explain.. Note that some time - a long time - after the ion grid, around 6-8" out from the main grid end (ion grid in the big tank, main in a sidearm) - we get pulses of fusion -top scope trace is the neutron detector (crappy circuit that has dc blocking issues, but the negative pulses are the neutrons). We had DC on the main grid, and an NST hooked to the ion grid via a capacitor - and you just can't drive a grid in there positive much unless you have a LOT of current. This circuit fries NSTs really fast, they don't like those positive pulses on the terminal when they're at peak negative in the cycle(!).

The timings look more like heavier particles, and yes, you're reading that right - the effective C and parallel R both go UP with density, to the point where as soon as we get anywhere near what most people think of a plasma density that would have ...waves, visible light, and so on, this effect is gone and it looks like a resistor to the grid terminal - with, at slow speeds, the usual negative resistance characteristic of a neon bulb. In other words, it starts acting like a more conventional plasma that might indeed have waves and stuff in it - if we went to those same pressures, we get those same results as the rest of the world - good thing in my view, I know my measurements aren't bunk.

I think we have something new here in the sense of working right between molecular and viscous flow, could be wrong but that's what it looks like - and I can't find any lit that works in this region, it's no man's land. Other measurements hint at ridiculously (ludicrously even for deuteron e/m) long transit times for what we think are the voltages involved, but have been hard to make and be sure of. In the dc + "put an impulse on there so you can measure times" mode, we're seeing at most 5kev on the deuterons by measuring speed - when we have 50kv applied. Sloooooow - and it does point to a more conventional interpretation of a plasma with waves and shielding and all that - or, looking at the geometry and the fact that we don't have the same density everywhere - not, hard to measure, this isn't a fluid in a tokomak, not even close. At 50kev if it were electrons I wouldn't be able to measure the transit times with my 2.5 ghz sampler! They'd be getting relativistic and my distances are short. At any rate, I did try to dupe the measurements I see in the lit for the plasma resonances and no dice in this pressure region - it's just not there. Shielding, dunno, probably yes from what I see, at least some.

I will have to crack the tank (which would be first time in ~3 years - it's REALLY CLEAN in there being pumped to e-7 or e-8 mbar between uses, continuously. What I need do to get those measurements is to have more probes at different distances from the thing I'm driving. Luck of the draw has it that they're almost all close
to the same radial distance, so doing the old distance/timeA-timeB = speed doesn't help me much - the distances are so close to the same I can't be sure of measuring them through the window by eyeball..no way I'd trust that. I'd rather have something like 3,7,20,25" distances or something like that. Which I could have in this tank - the main grid is in a side arm with the end about flush with a pretty big tank "drift space" if I want to use it like that.

In short, we have a far longer mean free path than works in the standard plasma equations stuff that assume hydrodynamic kinds of behavior.
Calculating from the Miley/Murali book1 for plasma resonance I get around 8 mhz and this should look like an inductance, but nope. At this pressure, "no mans land", there's just not enough interaction for that stuff to apply accurately, you can't assume a smooth fluid kinda thing.

FWIW, I'm saying indicated pressures as when in the lab, it's easy to write that down. Depending on which Pfeiffer document you look at - and here we're in no mans land between where the pirani and ion gages switch - the true pressure for unionized deuterium should be around 1/2 or 1/2.5 the reading. Complicating that is that already-ionized D reads a little higher...just an observation here, they don't mention that.


Now, at the low voltages, (3-4kv pp here) I don't expect this to look the same as at the "Real" voltages I plan to run to get velocities up to hopeful fusion levels. The reason for using this frequency band vs some other random one is that I should (heavy SWAG) be able to get to the tens of kv without hitting tank walls with D's - if I go faster, I can use more HV, slower and I hit the walls with less. Back to the low volts - at this scale, using the math I'm using (ion trap Mathieu stuff) the grid size itself is significant, which messes everything up somewhat. I want to get to 10x anyway of this so I know that's less of a factor and am working on modding a little Ham cw xmitter for that purpose now. It should also allow me to turn off the DC ion grid source, and use that as another faraday probe for timing measurements.
I could alternately drive that around 40 to about 80 khz but that adds a lotta monkey motion to account for in the other test results. Kind a high price for getting rid of the extra electrons, which I *think* higher drive on the main grid will (is) do anyway - and will do better when there aren't so many as the drive on the main grid along will keep it "lit". We have noticed, which I think is in favor of the more "plasma" interpretations - that medium HF takes a lot less to light off, or stay lit. This could just be the usual de-ionization times you see in say, thyratrons, but dunno, haven't checked, hours in the day and all that. We are at lots lower pressure than they so it should be longer...

It seems no math works right in this no man's land pressure range that we have "on average" and "outside the grid" and "away from the time/space of the focus region" which is why I'm doing this lab work in the first place.

We do note near-instantaneous response from the electrons when we try and drive the grid positive, it's a good diode to ground and fast enough that at this frequency range the phase lag is zero (as it should be). I will be taking more measurements of lag time to probe 6" away with drive voltage, all else the same, next run I hope - I need a better RF source, which is in progress. I also have huge, kW and multi kW stuff, but I'd like to work up to that before making too much $moke!
Next test will be with a mere 10 times the power of the last, a reasonable binary-type search, eh?

Probably the most interesting from my POV is that while my mass spectrometers, running at nearly the same speeds/feeds, barf out at around e-4 mbar, and here we're fine at e-2 - 100x higher pressure. Now, if I double that a time or two, this barfs out itself, but there's a pretty big space of ??? behavior in there. And right where I think I want it.

1 I've already just put that book back on the shelf as much of it turns out to be armchair theorizing that doesn't live long in the presence of test equipment, but if you like, the ISBN is: 978-1-4614-9337-2

Well, it's neither scientific nor polite to stick too stubbornly to a hypothesis, but nothing you've said indicates to me that it wouldn't be Landau damping. Landau damping doesn't require a hydrodynamic plasma, nor even a plasma with a thermal velocity distribution. It also works for nonneutral plasmas. You can derive it in the context of electrohydrodynamics, but it's actually much more general. (Sorry, Donovan, it's the Vlasov equation, not Navier-Stokes!) It actually works best in the long-mean-free-path regime you've got. Supposedly it even applies to the interaction of gravitational waves with galactic matter (i.e. stars as the particles)!

Ion trap dynamics is probably a good starting point, but as you point out, it's not a great fit because your density is probably at least a couple of orders of magnitude higher, so there are going to be collective effects in your machine that the ion trap analysis usually tries to ignore. The Vlasov equation should still work fine, though, with the caveat that your velocity distributions are not well-known.

I'm not saying it definitely is Landau damping, but seeing the presumably collisionless damping increase as you increase the plasma density in your plots is pretty suggestive. I still don't know why the resonant frequency would decrease as density increases, but maybe the resonant frequency is actually the difference frequency between the "tank circuit" resonance (excuse the pun) and the plasma frequency. You're probably right that the frequency is too low to be the electron plasma frequency, but maybe not if the density is as low as you say, and maybe it's the ion plasma frequency instead.

I'm willing to be educated for sure - it's the point of the exercise here.

I'd dropped further "conventional" analysis based on finding out that some of the books I had describing it (which mostly handled densities much higher than I have, and assumed "all ions" which I know is very false) made assumptions I'd proved incorrect in the lab already, and I mean laid them out flat and embalmed them - without even the preconceived notion that they existed, much less that they be wrong. Other assumptions seem wrong as well, at least under conditions I can easily produce, so..while right inside their domain, maybe not all-singing all-dancing rules -

If I'm reading this right (or if my own description is right - either will do), it should show me some interesting dependencies on drive voltage, as things are velocity dependent. https://en.wikipedia.org/wiki/Landau_damping

Which measurement I'm getting ready to take - my last attempt (on youtube) didn't have a decent time reference set up so a level triggering error in the scope cancelled out any effect if there had been one (I mentioned it at the time). That's not hard to fix at all, just wasn't something I could do in a couple seconds with the camera rolling and have a fun video. I'm only one guy out here....

I think we might be coming at the same thing, but orthogonally. For space charge waves to happen, things have to move, if only in the same way a dielectric like a water molecule aligns itself with an applied field and therefore increases the stored energy (via dielectric constant) by a mechanical mechanism.

I believe that at least one reason the mass specs crap out at e-4 mbar and we get to 100 times that is where we are running on the stability diagram - we're trying to hit the middle of utterly stable, and the mass spec hits just outside stable at all - they just wan the desired ions to last long enough to make it down the quadrupole and anything else be lost in only a couple cycles. We want them forever so are using a different U/V ratio entirely, along with the different geometry. But then at some few times our otherwise higher pressure, plain old mechanical collisions kill us. It makes sense with what I see on the test equipment.

I'm concerned with the next higher order - if I'm getting stuff to move, maybe I can move it where-when I want it (and now that it's moved, the charge distribution went with that motion, and thus back-reacts on the other stuff differently - you have to keep doing the perturbation, which I don't see handled there), and even have different regimes in the same tank at the same time - or more importantly, guide things so that at *some* times I have a lot of high velocity collisions at the focus and near high vacuum elsewhere, and in between, a more or less uniform distribution of things that are re-randomizing but while at very low energy as I've gotten my drive power back as things exited the grid. Whether that's what's happening now or not - it's what I'm trying to cause to happen.

I am getting collisions at the grid, I've tested that other ways (and seen the resulting neutrons - proof is pretty solid there). Now, to get them to happen how and when I want, while recovering the energy I invested in the ones that failed to fuse - that hoped for recirculation that does NOT happen in a DC fusor at all - extensive tests here came up nada. That smearing out you mentioned defeated almost all attempts to shock the thing with some impulse and measure the time that shock took to transit a few inches. It just disappeared, ditto any "ringing" we could find that wasn't just our stray L and C in the external wiring setup. Now I'm finally seeing transit times that are at least believable and what looks like a reactance that is, however, orders of magnitude different than "the books" and happening at frequencies an order of magnitude off from those predictions, and even as you pointed out, going in the wrong direction. I feel more like this could be one of those "hmm, that's funny" things, rather than just wrong. I'm real good with data aq and know when to be skeptical of my results, done it for decades in various fields.

Evidently Richard Hull was right about some things there. It seems most of the ionization in a DC fusor happens proximate to the grid, and the wrong end of the potential well...which would account for our measurements of very slow velocity of particles (on the order of microseconds across part of the tank - and very low effective energies). While, of course, the removed and secondary electrons that come from collisions with the grid do get full energy and bash the tank walls, as shown by where the heat flux and xrays happen - measured here. That makes some sense as the ionization cross section for electons on deuterium (and most other things) is greatest around 70 volts (count them) - at higher eV electrons short deBroglie lengths combined with short times in proximity just slip on by and have little effect on anything. In fact we did some tests of his theory and found it to be right, by having most all our D embedded in Pd/Ti sputtered onto the tank walls, so it'd be ionized there by the hot electrons hitting it, and ejected. That approach was interesting but had other serious issues re practicality - one is that heating let all the D go pretty fast - no way to keep that stable - another is that a heavy metal being hit by fast electrons REALLY cranked out the X rays a lot more than the stainless steel or sputtered Al we wound up with to suppress the spurious X rays.

So we really do have some stuff the regular models don't handle well here...I'm a big fan of the "standard model" in physics, BTW, just that I don't think a lot of the little add-on semi-regularities posited as laws hold always, particularly where not tested. I don't think we're going to need to posit any "new science" like the guys at CERN are looking for at all - we might need to notice some unexpected side effects of group behaviors that make total sense in hindsight, at most.

Remember I DO see more "effective capacity" at higher gas densities, and it's going up faster than linear (about .2 pf when I start to see any, and at around twice the gas, around 28pf)...something is afoot there, and more detailed measurements will have to be taken. As you said, this is NOT what the simpler theory predicts, yet there it is - I'm not cheating on this, maybe sloppy on some days at worst.

I'm thinking we can produce an effect analogous to "sound and compression waves" but without collisions where and when we don't want them, using electrostatic forces instead of the usual collisions. Due to geometry if nothing else, taking the gas in that pipe and pulling it into the inside of the grid is one heck of a "compression ratio", which is a poor analogy in one way, but an important effect to note for those used to analyzing things as if entropy always smoothed things out before I could get something useful to happen. I have a lot of peripheral hints that we do in fact have a compression ratio even with plain old DC fusor operation, and we showed some interesting pics of it with Bill's fusor just running helium - we could make things from the focus area flow out through a curved (insulating) pipe as if under what you'd call pressure, but things near the tank walls showed no such behavior, for example.


Now that they cut the cancer off me (the working fingers are crossed - the messed up one is still messed up) and my COPD is recovering, you can expect a lot more data in the near term. My tales of woe are largely behind me now - the roof is fixed, new solar infrastructure is in place, a lot of stuff like that I don't usually talk about here had me nearly useless for quite some time - now either fixed or well on the mend.

Hmm, the paragraph at the beginning of that wiki article is...ludicrously not even wrong in the analogy with a TWT...the external coil is kinda key, it's not just the electrons alone doing it.
And a klystron puts in a little RF (as we are here) and *bunches* the waves - the opposite effect described by the wiki, though one of the graphs shows how you do bunching, the opposite of dispersion (and we'll want that but it's easier with a more complex waveform than a sine when you only get one cycle/pass).

Someone might have been under the misimpression that this was a transmissive measurment - I'm not the worlds greatest communicator. Nope, this was the impedance on the driven electrode alone, not what happened after some distance out (like Landau stuff).

When I did grid vs probe out in the tank - I got the opposite effect, also - I got narrowing of the peaks! So, bunching, not dispersion. Probably does around the same thing as in a klystron drift tube - things go in and out of phase...I need more probes in there!

https://youtu.be/DKfnZM0VTlM?t=2m36s - on channel the drive, the other a faraday probe, maybe 5" away diagonally from the grid axis.

Here's two pix of screen grabs of that. The blue channel is what's on the grid itself. Note that this is DC coupled and not zero average - we are using a 750pf series capacitor here, and this draws
(more and with less transit time) current on positive peaks, thus putting charge on the cap and biasing the thing negative on average.

Going back and looking that this again (why do you think I take video?) I note that the yellow, probe trace is WAY off ground, and is also DC coupled from the faraday probe - it's off ground some amount that's about the same as it's peak to peak height (eg 1 pkpk + the ac too for 2 peak times the ac peak alone).
This also slid down seriously with only slight detuning - that wasn't me moving the vertical position at (whatever time that was - 1:50 or so).
Vert position is the little triangle at the left, same color as the trace. At one point I bring it off the bottom for the yellow one just to see how much *positive, not neutral* this plasma happens to be. With the DC ion source on (it's negative on another grid out in the tank) the whole mess goes negative on the faraday, but not on the main grid so much - but I should look closer than I did here. Sometimes we do this stuff just to notice what we've missed so we can go back and get it right.
So this is data that any hypothesis has to fit - at the volts here - about 3kv pk pk drive - electrons should be going considerably faster than this? They sure are at these volts in RF tubes.

I'll just leave this here as a nice example of something that needed no new science, but "whoda thunk?". Tony's always fun anyway.
https://www.youtube.com/watch?v=Hn8hDY4bvpI

Just some as-yet unanalyzed data I got today with a bit more powerful RF source (around 30w).
The pic is a long exposure - this is barely visible to the eye. The scope is that same xy but I reversed the channels. Note additional notch in the faraday signal. On the scales, the faraday probe is that 2.5ghz quarter wave antenna used as a probe, loaded with 100k and the scope screen volts are correct.
On the HV drive you multiply the scope by 9 to get the right answer. You can bet my proximate Ne2H bulb nearly exploded at these levels. Couldn't really get reflected impedance with this setup - it's what I bought that analog discovery thingie for, but haven't lashed that up yet. Since I have to keep tuning the RF source to keep stuff from melting, it's always zero phase at that spot - I can do a better test with the AD 2 as it has more inputs to look at stuff with and should be "fast enough".

I'm looking at things like 1b3 or 6bk4 to be able to adjust the amount the grid is allowed to go positive...the equations I'm using seem to think the ratio of AC to DC matters...and it seems once I've sucked out most of the electrons, it stops being such a good diode itself. This will likely be a sensitive funcion of ion density as well, but too many variables make it hard to get interpretable data.