Interesting indeed, hopefully not as in the Chinese curse...
I'm currently driving the H bridge with an IRS 2453 - it's about the simplest H bridge driver out there, but it has some limitations on speed and gate drive current. I'm looking to put together something a little more sophisticated soon. This setup is near the limit on speed as is, already wasting some input power due to slow turn on and off times.
You pretty much called out the issues on the stepup transformer situation...but for research, we can force it some (more than I thought at first). Once we know what we want, I can do a redesign and hit the numbers a lot better, but anything with this turns ratio isn't going to do octaves of bandwidth decently - and for the arbitrary waveform I suspect we'll need in the end, it'll probably take two in series, one for the fast part and one for the slower part.
As is, it only has to be efficient enough to not go up in smoke in the time it takes to get data, which can be pretty fast - and pulsed operation is a side benefit of all the remote control stuff I've put in.
As to interpreting that data...a few things to keep in mind. And don't forget, I'm guessing too - I might be on a better perch to do it, but that just increases the stress to get it right.
Things to note:
1. The geometry of the tank. What we want eventually is for all the ions to be inside that side arm pipe, stay in there, recirculate through the focus, and not bash the tank walls while outside the grid.
2. The Faraday probe isn't in the sidearm, and is far enough away that if it sees anything, it might mean we failed #1 - or just that the fields out there are weaker so things flew further in that direction.
3. The probe doesn't say a thing about how far a burst of ions was away at nearest approach (other than how loud the signal was) or how many (also how loud the signal was) or what direction - no indication at all. It's pretty limited information. That said, no one else ever even bothered to get that much.
4. The neutron detector, while time accurate and low latency, is deaf as a hammer. Calibration is a million neutrons a second gives us 980 counts per *minute*, Numb. But the ones we get we have very high confidence in.
The Matheiu math describes what you need to put on some electrodes, a mix of AC and DC, to restrain charged particles (ideally, just one at a time) inside some limits - eg stay in the trap or between the electrodes of a quadrupole mass spectrometer. The "just one" is because even that fairly hairy math (to me anyway) doesn't take into account any mutual repulsion of the like charged particles, and of course takes no account of any opposite charged paricles that might be flying around (in our case, electrons mostly). This stinks, but it's the only math that even addresses what we have at all, it's just going to need some more terms to handle that stuff. Know any super hot math guys we could hire? What I did discover just now is that it's not as bad, or at least all the time and everywhere, as I'd feared. For most of the cycle things are so not-dense the original math is pretty close, or should be based on some other measurements of transit times and trajectories I made earlier. Whew....Here's a little background that might seem a bit off topic at first, but the good stuff is in this very expensive book Bill bought and it'd take a long time to scan all that in with nice pix of trajectories and so on. This is a start, anyway:
https://en.wikipedia.org/wiki/Quadrupole_ion_trapNormally in an ion trap, the residual small forces between the few ions tend to drive it toward an end state where they keep maximum distance between themselves. That's not what we want, but that equilibrium takes time, and if we can get this done on the first few cycles...that won't matter. There's more - a lot, but I only have ten fingers and this much time. For short, when things are slow and low energy, near the tank walls, if they get uniform due to mutual repulsion - we've already gotten our energy back while slowing them down, and this creates a nice uniform start for the next cycle without costing us in energy investment. (when the grid is biased to slow things down, we get energy back into the power supply!).
Now, why did we see what we saw? Well, I have a theory, let's call it a guess.
Remember, most of the time all we see is stuff flying past the faraday probe (lower left in the picture) at some point after the grid has passed its negative peak. We'd expect that even if we just drove it negative and it sat there, though the timing would be different. After being attracted in from all over the tank, they'd go through focus and go back out the other side, slowing down, but...eventually going out as far as they came in from, more or less.
Having the grid go positive while they're flying away would just make that happen faster.
Now, suppose that with the alternating grid potential, we have the ions doing big old loops through the big part of the tank - we don't have it "right" to keep them in the sidearm, and even if we did, there's that open end...
After some of this, some will find a happy place - a distance or loop size - where they get into near sync with the drive signal. Call it "selection from random data" - which is dreaded when scientists use it to cheat, but here it's jut that the ones that don't hit the tank walls and are lost - so we only see the lucky ones. BUT! There's more than one set of speed-size parameters that would accomplish this, and in this data we see up to around that aren't always in phase. The other two in this sense might just be from previous roundy-rounds, we have no way of knowing. (in guessing, it's important to know what you don't know - known unknowns and all that).
As an example of abusing that one - I noticed that when we have the multiple peaks they aren't as large as when we had the one...which could mean that the total ion current has divided itself up into these guessed-at loops.
Or it could mean anything, actually. We have no way to know from this data. Rigor isn't as much fun when guessing, but it gets you there quicker, usually
.
But this is of course, just guessing - hopefully limited to the things that are possible, but still, a lot of things are possible. I do note that all those with the multiple peaks were seen while triggering on that numb neutron detector - there could have been neutrons on every drive cycle (I assume there were, it's fairly safe, the thing is truly numb) - and when we free-ran - we didn't see neutrons on the scope, there weren't common enough, nor were the multiple peaks. Just didn't see them, and yes we looked.
So, what could have made that come and go? Those dreaded missing terms in what will be the real math describing this is my best guess. We have a lot of stuff sloshing around in there - it's not pure D+ or even D+ and e-.
We have D
2, D
2+ D and D
2- (yup - measured that once), and various contaminants sloshing around in there. It's a ten gallon reservoir with bumps and warts and...the part we're interested in is 6" diameter and 6" long hanging off one side - with an end open to all the rest.. The Mathieu math only gives confinement and re-circulation for one charge to mass ratio at a time...and there's about zero chance we hit the right set of numbers here. In looking at the plots in the expensive book on Quadrupole Mass Spectrometry, some of the orbit types and shapes for off resonance particles would easily account for what we saw. It's my best guess at this moment, anyway.
Here's a few scans from Peter Dawson's "Quadrupole Mass Spectrometry and its Applications" so you can see what we're dealing with, kind of. It's the only extant math for this. Yes, what he describes is a different geometry (but the same math works for a few he does describe) and under somewhat different conditions - but for part of our cycles they are about the same - but it's all we really have to go on, it's just the implications of the equations of motion (Newton) for charged particles in a field imposed externally. It does NOT take into account the fields created by the particles themselves - that's the extension we have to make. And it'd be nice to have it feedforward as in "what do I do to get a result" rather than feedback "this is the result if you do this" form, but hey. If it was easy someone else would have done it. And if the conditions were common, we'd see it out there in the sky with our telescopes. (Feynmann and others).
Note one thing not obvious to people who haven't spent a lifetime looking at scopes. A low frequency thing - even just a delay (due to a path time) can look just like something high frequency thing if the delay causes a "re-play" of the high frequency thing later on, perhaps almost on top of the other high frequency thing. I suspect strongly that's what we're seeing - a pulse from a previous roundy round whose trajectory eventually brought it back by the sensor.
Note in these scans that the X and Y are synthetic variables that take into account amplitude and frequency of the AC and amplitude of the DC. The search space we're interested in is that plot that looks like a spider - the intersections are where things go round and round and don't hit the wall.
(sorry the board doesn't seem to want to preview gifs today - click on them)
- The inputs
- the spider
- Some possible trajectories
This will all of course be a little different for us - our geometry and goals are different. But the equations of motion are what they are. Pretty scary stuff, juggling charged particles with an attempt at applying fields that will always be imperfect, and affected by the tank, and by the particles themselves - and in our case, the density of the particles changes as they go from diffuse out in the tank where this math is pretty close, to compressed (yeah, we have a compression ratio, but that's not what people think) at the focus - when things are at the focus, which isn't all the time.
Further, what I'm pretty sure we want - and I was surprised that a simple sine did bunching that well - that was amazing - is a slow onset to negative, but a quick reversal to positive when the ions are inside the grid, then back to negative to slow them down again before they hit the tank walls and are lost (they cost energy to create...we want our nickel's worth here).
The idea behind the reversal is to drive them in further, but mainly to draw in the much faster moving electrons to get between the ions and promote fusion, much like a meson would do. At some relative speed, they become effectively "small" compared to a deuterium nucleus wave-function...which should allow some of the Coulomb force to be neutralized and allow lingering long enough for tunneling into fusion to happen...that's my theory.
It pays to know that other things being equal, electrons move ~60 times as fast in the same electrostatic field, so we can hustle them around before a deuterium ion notices much. This is the square root of the charge/mass ratio ratio. If that makes sense.
Since strong force kinds of things are 10
20 or
22 per second, and this is more like 10
-9 of a second near focus...well, that's a relative eternity for luck to give us fusion, eh? Both are fast, but you have to have a sense of proportion (H/T Hitchhiker's guide to the galaxy),