I hypothesized that running a fusor kind of like an ion trap (using some of the same math to predict the electrode drive parameters) would be an interesting thing to try. This essentially also means that there is recirculation, and unlike theories where that somehow happens by magic, this involves actively driving it and doing so in a way that avoids things bashing the tank walls as much (or ions maybe at all), and giving things a lot more chances to collide than would normally be the case.
While this run didn't make record neutrons - I deliberately set conditions near the threshold of "any at all" so it'd be sensitive to parameter changes - it did quite well on my arbitrary scale of Q (I haven't normalized my graph to real watts out / watts in, as it needs to be big enough to see!).
Now, a mass spectrometer tends to run out of steam somewhere around e-4 millibar or lower - maybe e-5. This is due to two things - the math does NOT account for any interaction of the particles with one another, and they generate their own fields that distort the one you think you're putting on there - and because once there are a good number (shorter mean free path) - they collide and knock themselves out of the quadrupole. I'd add that mass specs are run on a parameter line that is just barely stable for long enough for the ions to make it down the quadrupole - or not - the object there being to mostly reject ions that aren't at the e/m ratio being measured. In normal ion traps that are run closer to the stability center (but not much) - after a few milliseconds, all the ions have "learned" to avoid collisions and you've got this pretty picture of the recirculation dance - but no collisions. However, milliseconds or seconds are a really BIG number of passes at multi MHz frequencies, and up till then, they do collide....which is what I'm going for here.
We have a somewhat different goal than what others have been doing with this kind of drive and electrode setup and some different rules would seem to apply.
And in fact, what I've just seen is pretty much expected - under the high vacuum conditions, long mean free path, the existing math works! More importantly for me, it means I've understood it well enough to apply it to this rather odd electrode arrangement and these particular ions, so I have a baseline to look into adding the terms anyone can see we're going to need to account for the particles effects on one another and the net field, which will now be really different than the applied one (which we already kind of knew from the DC fusor operation - most ions were created at the bottom of the E field, and measuring that gradient confirmed that the net field was nothing like what we applied).
So, I did a run, starting on the numbers that some simple script I wrote predicted ought to give interesting behavior, and then twiddled all the knobs while still taking data to see what I could find, and here are the results of that.
IT's going to be obvious that I need to add a few columns to the database (actually, that's not going to be hard, just take some time), and do some more runs even with what I have, but getting some more scope traces of where the ions are when - phase info, bunching, spreading, whatever, vs the parameters. What IS encouraging is that even at around 2 orders of magnitude more pressure than mass specs crap out at - this seems to be a valid approach. I'd go lower, but right now I don't have an ion source in there that will go lower...another thing to add - all these answers, as usual, generate more questions or at least the need to measure more things (same?). For this run, the scopes were set up to give me early detection of things that might catch fire, or fry me if there was an accidental resonance that made the fusion work really well - it's happened, and I like staying alive. Now that I have more of a feel, I can move the resources around more appropriately.
Now, there's things on this graph that make more sense if you were there and knew when a knob was being turned and which one. For the moment, you're going to have to trust me, or not.
What to look at here is the Q at the 25kv line vs current, which is really pressure, I need a better color map there, but maybe obviously, current goes up with pressure when all else is equal (and here, it is). Note the flat line at the bottom of Q - that's without adding the RF. Then see the very fast rise at lower pressure with the RF (which was around 8kv peak at 2.91 MHz in series with the DC).
- Q at a range of conditions
I got similar but worse results in Q at the 35kv line - which the math predicts, and which is where the bypass capacitors in my matching pi network crap out. They ar needed to make for an RF ground where the DC is fed in - I don't want to fry that nice Spellman SL2KW.
I have a bunch more plots...it's nice to have this in database, including some vs time and with comments on when turned things on and off...or changed something.
Available on request. Right now it's apparent to me that I need to make a better color map with bigger shifts around the pressure I'm running at, and another preset that maps that axis to time instead, or adds little arrows (yep, that's an option in gnuplot) that shows the order in which those scatter plot points happened.
But maybe even more important is to take a bunch more data around that sweet spot I've identified with the help of the ion trap math, and look for what changes and how much right around there - get those other scope traces utilized to see bunching, spreading, phase shift and what not else, I do have a way of recording those as well, it's just usually too boring to bother posting - minutes of more or less the same thing with variances only the trained eye would know are significant, and a little bit hard to get synced up with the audio recording I'm also making on another machine; after that one wild event, I'm trying to capture everything I possibly can in case something else that's evidently improbable but possible happens, so I can replicate it. But the effort of putting together a really full report isn't worth it for every single run.
Right now I'm kind of hampered by the 6 hour or so time constant of indoor temperature of the lab with both a woodstove and a propane heater cranking, combined with low winter solar power availability. This isn't the best time of year for this...and I'm a bit of a wimp when my fingers freeze to the knobs.
But it's something - and the breakthrough here if any is that the math I've managed to extract and munge from various sources does seem to work, so I now have a fairly solid starting place to develop the real thing that will be more predictive for our special set of required operating conditions....eg a lot more ions/cc and more collisions. This particular excercise was a test of a few numbers in the a/q stability terms I kinda picked as "brown numbers" - and I got lucky. Brown numbers are those you kinda reach behind you and down in your pants to fetch, and hope for the best. So hopefully I won't have to repeat that explanation of a "scientific wild-ass guess".
Here's a little background on that. It's not actually what I used, which was a far more expensive book by Dawson (thanks, BillF), but I can share this.
Most mass specs run on a line that is kind of a poor fit of one edge of region "a" in this. I went for the middle. Since a and q (their definition of q) are reduced variables that really are ratios of this and that, it's possible to hit a particular region at either low volts and low frequency, or higher for both and I went for the region where we just start to see neutrons out of the noise with pure DC as for this, I kinda have to be in the lab with it to adjust things and didn't really want a huge output anyway...but those are basically free parameters that can be moved around once a fuller understanding is had....
Obviously there are going to have to be more terms in the basic math, or some what of accounting for the particles' effect on one another and other smearing, and that's what I'm trying to get a feel for right now - then we can really go places and do things.
Using-Graphical-Tools-to-Understand-Quadrupole-Theory.pdf- Someones simplified paper, a little helpful.
- (741.32 KiB) Downloaded 350 times
