Hello from .au

[Remy Dyer]

Greetings all!
Apologies in advance for the wall of text to follow. TL;DR version = (30 / M / Computer engineer, graduated 2011)

I'm a computer engineer by trade - currently working flat out at CSIRO's Energy Technology Centre. I'm still partly fresh out of uni, although I did take my sweet time finishing my degree, and actually ended up landing my current job before graduating. I think I'm OK at what I do, but have come far enough to appreciate the difference between theory and practise. It's humbling to learn things the hard way that one thought one already knew. (most recently ADC aliasing, and how just about everyone seems to get their antialias filters just plain wrong, notably including certain embedded control equipment manufacturers. )

I'm working for a team intending to have the last word on Carbon combustion efficiency. Hardly glamourous, and I think not at all the answer in the long run, but it's a good learning experience and foot in the door. The premise is an old one - run a diesel engine on coal, get slightly higher potential efficiency, but more importantly, get better load-following performance to enable other, more unreliable energy sources. (Particularly renewables). Well, it works, and we're moving on to the full scale fuel system (big slow two stroke, the kind that redline under 200 rpm). Probably still a few years yet from commercialisation, and given Hofstadter's Law, likely more than that yet.

I'm the lab monkey responsible for the control systems as well as the data logging and analysis. Mostly working with National Instrument's LabView and compact-RIO embedded systems (although, if I were going to do it again, differently, I'd steer clear of it and use only tools I could see the source code for, nothing more frustrating then trying to second guess a damn black box). I'm also not shy about sticking my nose into problems that I'm not qualified on paper to handle. At the moment this has finally gotten all the way to the rocket science. (convergent-divergent nozzle design). But has already included non-newtonian viscosity, as well as second-order (read: very fast) hydraulic solenoid valve control.

My passion, however, is fusion. Particularly Bussard's design. I've been more or less obsessed with the idea since I saw Bussard's seminal talk. I'd really like to head a team here to look into IEC fusion, although this is actually quite a tall order.

You see, when CSIRO was first formed in the '60's as a consolidation of government funded research, all work into nuclear energy was dropped. The original CSIRO charter explicitly excluded any such, and the other nuclear science was separated into ANSTO. (Whose remit does not include "energy technology". Although they do some interesting things with neutrons; notably the invention of non-destructive neutron activation spectroscopy.)

However, in 2008 this was quietly fixed, the new charter now talks only of "freedom to peruse..." and does not mention "nuclear" at all, although now there is no-one at all in the department "qualified" in anything nuclear. A bit of a chicken and egg problem. Looking into polywell/iec, the first thing that comes up is Nevin's '95 paper - and this is enough for most not in the field to consider it "debunked".
Well sadly "debunked" is a harder stain to wash off than just about anything, and if you still believe it might work, automatically you're a crackpot. I might yet get this some serious funding, but it's a damn long shot. It doesn't help that the site as yet has no radiation safety officers. (pretty much all "nuclear" work is done at Lucas Heights, generally under ANSTO, although CSIRO has close ties).

Well anyway I'm very small fry, and I'll generally try not to call out my CSIRO affiliation on these boards. We actually have fairly strict rules about that - even clients using tech we've commercialised for them are not allowed to mention CSIRO on their product labelling.

Just to be clear, I'm not hiding secret government "mib" status, nor am i going to run off with anyone's ideas. My opinions are my own, and not in general held by CSIRO, nor do I speak here with any of CSIRO's authority.

And that last is why I don't have "CSIRO Research Projects Officer" on my "occupation" tag or in my 'sig, despite how impressive it sounds. (personal opinion: CSIRO's "unintentional submarine patent attack" re: wifi was a bad thing to do. I think Patents ought not to exist. Maybe demand all mass produced products file full analysis design documents, so that loss of the company doesn't equal loss of the knowledge. Would also make paperwork for the "patent office" people to handle.)

Anyhow, personal projects wise, I'm into audio gear in a little way, mostly into improving sound performance by pre filtering audio to suit the speakers. In particular, so as to have a flat group delay. (my hypothesis there is that we can actually hear very small phase delay information - it's what yields most of our "directional" sense of hearing.). Within which I've had very good success running embedded linux systems to run the filters on. I assemble my own amp's etc.

My long term goal is to get a physically full scale polywell demo reactor running. I want to go straight to full physical scale, since performance is (supposedly) determined by the strong scaling with radius. One doesn't simply get to play with a small scale device, with appropriately scaled down electrons, they are what they are. I'm going to go out on a limb and assume that Bussard wasn't senile, and was rightfully confident that "all the science has been done...". The key problem is, as always, cost. If you don't get to choose the scale, and you do it in the easiest way, then the cost is considerable. ($200M '05 USD?). My plan here is simple - cheat like crazy. Avoid off-the-shelf parts (excepting those from large-chamber EBM machines). Opensource hardware, and make very big turbo molecular pumps, or even use very many sprengel pump tubes about the periphery with a single large pump to recirculate the pumping fluid.

My gut feel for the physics is that self-sustaining thermal distribution energy chain reaction fusion is just not going to happen short of something with the physical size and plasma density of a small star. (do you suppose you'd see planet size stars if this wasn't the case?). My feel is with a thermal distribution of energy, at the high density you need for self sustenance, brem is always going to beat you. (as in Todd Rider's pessimistic analysis). However I like to believe it's still possible to do, but you need to spend considerable power to keep your energy densities non-maxwellian. Here is where I think the problem is: Fusion researchers have, for the last 50 years or so, never studied the gaseous electronics that those who wrote the textbooks knew. Following Bussard's lament about no suitably trained engineers, I tracked myself down some books on just that.

I have also ended up getting into alternate electrodynamics theory as well. I like Phipp's work, and have had much the same beef personally with Maxwell's Equations as I suspect many engineers over the years have. (Spoiler: they're just an approximation useful for antenna design, or for particle accelerators where there is only insignificant charge moving vs the coil fields. Use Ampere's or Faraday's laws directly if you have to seriously design/analyse electric motors or solenoids, the total derivate is important.). I particularly like Phipp's "Weberian" force law. (Spoiler: Einstein was wrong, there's certainly time dilation, but no space contraction, and any clock is ok, so long as it's compensated properly. This is because the time dilation is actually reciprocal, as is proven by GPS operational data. No big conspiracy, just a lot of incompetent people who like to pretend they're not.)

I'm looking into checking it out as Wolfram suggests with his "A new Kind of Science" book, i.e., exhaustively by large scale simulation. I can use the aussie supercomputers for this, I have access and speculative work is OK, the billing is retroactive and only if it's useful to a client).

Most of this last stuff I don't do at work, nor to I leave much lying around. IP issue again, but they're not paying me to theorise. (yet). RPO is kind of a high grade of support staff, rather than a low grade of "research" staff. But I have a cool boss. Will likely change after I get myself a PhD of some description, probably on the CFD fluid mechanics stuff, which is of course very closely related to electrodynamics.

Anyhow, to any who've followed this far, you have my gratitude, and I'd be happy to court flamewars, but they should be placed in the appropriate thread.

As I am still a lot more talk than walk, certainly feel free to disregard me at your leisure, I don't mind
I am still going to speak out where I feel I know something, even if I'm proven wrong. I have pretty much chosen this board to do so thereon. (Was considering the talk-polywell board for quite a while, but S/N is a little high, maybe later.).

What drew me here was the practical nature, and the desire to document the mistakes and problems. Not nearly enough of that goes on in "serious" published science these days.

Experience is measured by equipment failure ^_^

[Doug Coulter]

First welcome, and nope, no problem with walls of text - I could use the competition...

I wish we had more "it almost worked" posts up here, but it seems psychologically difficult for people to admit they went for something that "seemed like a good idea at the time" without more homework up front.
It's just hard for some to admit failure, even though it's by far the most instructive thing one can give back to the community, I think - it saves the others from going down wrong paths, and identifies the gotchas that all too many ignore in their thinking.

I haven't seen the paper that "debunked" Bussard, but I did debunk it myself - along with the way most people think farnsworth fusors work - they're dead wrong and I have the measurements to prove that, in spades.
I've measured recirculation - it's nil, not even a little ringing on a ghz scope trace. There is not a neutral plasma or mainly nuclei - it's electrons in bulk, about 10x the number of positive ions.
The fusion isn't even mainly happening at the "focus" either, as both Tyler and I have measured some localization out at the tank walls - from the very few negative ions that get produced slamming into others embedded in the tank walls. It's pretty dramatic when measured various ways. One thing I did to prove that was to coat the tank walls with Ti/Pd to increase the number of D atoms there - and it doubled my output instantly, at least until they heat up enough to disgorge the fuel (which you can also see on a pressure or volt/current trace). I plan to write up the fusor "misconceptions" at some point when I have so much convincing data there's no possibility of argument about it.

I'm not saying you shouldn't try a polywell - you'd be the first if you actually get any neutrons. Funny that Bussard didn't ever really prove he did - even with government money he somehow lacked a realtime sensitive neutron detector that every amateur fusor hobbyist has - I find that one extremely suspicious. And that outfit that says if you just give them a few hundred million and they'll make you a Bussard polywell power plant - even though they have no prototype, and no results - that's gotta be one of the many scammers in this game ( I do believe that some of the scammers don't realize what they are, a starry eyed blind belief can have the same results) - and things like that are one reason my own group doesn't go trolling for dollars - we don't want to be painted with that brush in any way, like so many others have (deservedly) - in and out of "official big science".

My take would be to forget "full scale" at first. I'd completely forget the idea of a chain reaction in a fusion situation - plasma of any real density doesn't absorb the products of the reaction and pick up any significant energy that way, even in the sun really, which is the WORST fusion reactor in existence (thank god, we'd be vapor). Do the math on that, and don't do the stupid obvious mistakes of ommision so many farnsworth fans do (like particles are only charged when that's handy, but neutral when that's handy - in reality, they are what they are, handy or not, at a given instant). By what mechanism can you get the gamma rays re-absorbed in a plasma? That's a really hard one. Or neutrons...work it out, it's horrible numbers for a chain reaction, and we don't want one anyway, really - all we need is more out than in, and we can feed energy back in to keep a reaction going if we've got it.

What I think you're going to find if you try is that the same things that make electrons apparently easy to control is also what makes them almost impossible to control - their tiny mass (or big charge/mass ratio) and they are the big loss factor in all this as far as my own readings go. Also, the idea that you can make his proposed magnetic field is bogus - you need a monopole with holes in it to get what he thinks he can get, and I don't see that - you just move the leaks around with real magnets, not eliminate them. I think the idea of using super-light electrons (compared to D nuclei) to attract the nuclei is insane - if a light thing tugs on a heavy thing, which one move the most? How does that play in with the equations for the magnetic effect on the electrons? And their self repulsion...working on a much lighter mass...it just does not make sense in any obvious way, though I've been surprised before - by actually working models of things, which the polywell apparently has zero examples of so far. If anyone has one working, they've kept it really secret - there's no data out there I can find.

Again, that said - try it, man. Do it on some scale you can afford, because you're going to find out (ah youth) that there are a ton of icky details you have to learn first that aren't really scale dependent, and you're going to "waste" some time and effort learning them all. Better to do that with smaller amounts of money and gear, as you learn the same either way.

I've talked to the main simulation guys (SIMION and others) and nope, your supercomputer ain't gonna get it - their software won't handle it anyway, and doesn't scale at this point to a many-processor or huge-memory situation anyway. You'd have to write it. They can't handle all of the aspects of even a very small subset of the number of particles involved - you can do RF, but not in the presence of AC or DC mag fields, and junk like that. They kindly warned me off paying a ton of money for their software when I told them what I expected to do with it. If you try to "cheat" and use very few atoms so they can handle it, they still can't handle the scales (tiny atoms, big chamber) and there's this little detail that with the wrong amount of atoms, the mean free paths for collisions are all wrong, and the behavior changes qualitatively as well as quantitatively. This really is an "all or nothing" issue.

Re our private discussion - I think I'll start a thread on metrology to discuss those issues, as it would be good for other people to see too. I've spent a really long engineering career doing data aq, so I do know the issues there, no sweat (I'm a published author on DSP technique as well). So we can have a go over there and educate everyone who doesn't already know this stuff, which I assume is quite a number of people from the questions and assumptions I see going around that "aren't even wrong". FWIW, yes, I have looked at the time dependent behavior of Fusors in various ways with gear that handles into the GHZ and have a few things to say about that I've omitted so far, but there are indeed some interesting phenomena even at much more reasonable speeds too - it's not all DC. However, some basic data logging - even with small resolution (12 bits @ 1 second) is better than most people now have on their test gear - and it's affordable and reasonable for an inexpensive product we could make for the community, with the right specialization for this work - so we are doing it, knowing that it's also nice to have a really expensive DSO with multi GHZ sample rate to get screen captures off as well - but that most can't afford to add a few k$ to their home experiment budget at the moment. I am really NOT a fan of using delta/sigma soundcard a/d's for much of any of this work, the ringing from the sparse-tap FIR filter in the front is too nasty, and the artifacts of most approaches render the data almost meaningless. I DO use a vid cam to get audio/visual annotations of the more interesting things, BTW, and you can find some of that here:
http://www.youtube.com/user/DCFusor
Though there's not much need to "go direct" as most of that also appears here on the board, and here is where I add the explanatory verbiage that makes the videos make sense.

[Remy Dyer]

I'll take the "agree to disagree" bit re: polywell, merits discussion in another thread.

I agree with pretty much everything you've said there, especially the part about the sigma-delta converters meant for audio.

About chain-reacting fusion - I didn't word it well. Break-even is what I like to think could work, but never with a chemical or fission style chain reaction - I agree with you 100% there, it's too thin! (Ditto heating with gamma).

I have a pet theory about why internal combustion engines get so little efficiency - and it goes something like this:

Flame is a plasma, and hence conductive. So as a conductor, it absorbs and emits light. Being the hottest thing in a combustion chamber, mostly it emits light, rather than absorbing it. Attempts to improve efficiency by insulating or polishing the inside of the chamber don't make a lot of difference, because the colder working gas is transparent. That light, at flame temp, eventually gets absorbed into the combustion chamber walls, which are the coldest thing in the chamber. Hence that energy gets to jump from flame temp ~ 2000K or higher, straight to hot metal temp at ~ 423K, entirely bypassing the gas in between. Having never heated that gas, it's impossible to obtain mechanical work from that heat flow.

I did some measurements at work which pretty much bore this out. Those measurements were in large part only possible because I got my ADC gear to actually sample at 20 kHz sample rate with 16 bits of actual precision. (i.e., noise floor truly down around -95 dB, and absolutely no aliased noise).

Some guestimation about the size of the flame surface, the knowledge of the fuel quantity, combustion temperature and energy content, along with the tiny spike from the combustion heating gas in our "constant volume spray chamber". (it's named HECTA : Hi Efficiency Carbon fuel Test Apparatus) and stefan's law actually yielded a power balance that seems legit.

Anyway, combustion light in an IC engine is akin to bremsstrahlung in a fusion reactor - it can only be a loss channel. It zooming through the gas or plasma is not going to heat said gas or plasma. For more or less the same reason - transparency. Hell, even tungsten is "transparent" so much as gamma is concerned. It may deign to scatter, perhaps, but otherwise it's going to ignore such puny average density.

The paper I meant is http://fsl.ne.uiuc.edu/IEC/Nevins,%20Ph ... (1995).pdf ; The relevant quote is

(...)Hence, we conclude that inertial electrostatic confinement shows little promise as a basis for the development of commercial electrical power plants.

I usually parse "recirculation" to be the power flow that has to be injected to keep it going - it's the energy you press in with your high tension supply, not any "Flow" within the plasma.

BTW I also agree about the "plasma" thing. Fusor tubes don't "contain plasma" or operate by the purely thermodynamic rules of plasma physics. Such rules only apply when the approximation of quasi neutrality is a good one, otherwise what you have is gaseous electronics, dominated by space charge effects and other electromagnetic interactions. Which one doesn't get to "pick and choose" like Nevins does. If you've got flow of charge, you have magnetic effects, if you have space charge, you have electrostatic forces too. Most of the kinetic gas theory of plasma is an approximation that applies where there's no "space charge" because of quasi neutrality, and the theory has a blind spot to magnetic interactions too - mostly only considering externally generated fields and ignoring fields generated by the flow of charge within the plasma as insignificant.

This point of view totally ignores Lenz's law, so of course you end up with turbulence and magnetic reconnection and electrohydrodynamic instability and so on. His work *could* be done right, but sadly n-body relativistic electrodynamics is an open problem in mainstream science. Relativity denies it.

Ignoring relativity gives computable simulations using a reasonable number of "super particles" which seems to work. (where a reasonable number is ~ 16000). Sure this is still many orders of magnitude away from the real number, but sim still seems to capture the dynamic effects of the big mass difference fairly well. The guy doing those sims eventually gave up, with the conclusion that the wiffleball effect that pinches off the cusps in a polywell was real, and would work. There's a reasonable basis for ignoring relativity - Phipps' books "new physics for old" more or less sums it up. In short, fix Maxwell's equations plus "force law" to be properly invariant, rather than merely "covariant" and over specified. You end up getting a few interesting outcomes, one of them being that although you still observe time dilation, if you don't care what proper time the particles see, then you can just pick any clock and get on with your many body sim just fine. Oh, and also no length contraction, as time ends up totally invariant to any spacial direction, regardless of velocity.

In short the fur ball of electrons free to move as they liked had a strong tendency to null out the cusp field within the "wiffle ball", just as do superconductors and for similar reason. So at the wiffle ball edge wherever you aren't near a cusp "hole" the field direction is tangential, and the gradient of the strength quite sharp, so that particularly energetic electrons going for the magrid end up bouncing back in, until they try slowly enough to penetrate to the coil container.

My specific complaint about Nevins, is that he ignores this magnetic interaction too.

And Bussard did have a neutron counter - he had two, but he apparently never used any analog to digital gear (he mentions having big problems with electrical interference, specifically with the neutron counters) it was all put on paper with one of those rolling paper plotter machines. Damn, can't think what they're called, but could lay hands on a few at work in storage...

Anyway, one of the published reports he put out had a picture of that trace.
In all of the six or so runs before WB-6 blew up, there was a count on the "near" counter in exactly the 1/4 ms interval that the polywell was in wiffleball balance - which he couldn't hold it in, firstly as the magnetic field and electron gun drives were just control by firing them off and letting them go. This was no problem, as it turned out, his theory predicted that any such curve of relative gun and magrid currents would always pass through that balance somewhere in the discharge curve of the cap bank, regardless of whether the electrons or the coil made the stronger magnetic field first.
He did lament this a bit - doing things this way meant burning up most of the cap bank on getting enough electrons in there for them to stop leaking out - they had to be driven damn hard to start with.

Secondly as he didn't have a nice fast solenoid to puff the fuel D in, all the runs stopped with the expanding wave of neutral gas causing Paschen arc discharge to the walls when it got outside the magrid.

He had a "far" neutron counter across the room to show background in the room. It was around 20 or 30 odd counts per minute, from memory. The idea was it was far enough away from the reactor that the flux it would get from the reactor, assuming all the neutrons were coming from a point source near the centre, would fall off fast by 1/r^2, and so it would be very unlikely to count any from the actual run at all. The idea was if the counts were actually just electrical interference, that it would count at the same moment.

One of the runs even produced a double count. Again, at the moment his theory predicted the electron current into the polywell would be maintaining enough energetic electronics within the grid to be in the "beta = 1" sweet spot.

I know it's not "proof" - what, six, seven neutron counts? All of those counts could have happened to be background coincidence, at exactly the right 250us interval, every time they ran it.

Oh also - that beta is not what most plasma physicists think when they hear beta - their beta includes the force on the ions, which is not the point - bustards beta is only the electrons vs the magrid field. I have seen arguments that beta =1 can't be reached, and they're largely right - the difference in mass between ions and electrons, means that ions are affected a hell of a lot less per charge than the electrons, so the field you need to "balance their kinetic pressure" is the same ratio higher. Ergo, tokamaks with super fields, and the approximation that the fields from the particles are insignificant against those giant superconductors...


Most of the interesting papers he actually published seem to be all typewriter printed, by dictation. And what's in digital format seems to have been typeset only with word - bad news side by side with Nevins, who at least used LaTeX, from the looks of it. Which one looks legit? http://www.askmar.com/ConferenceNotes/2 ... 0Paper.pdf

I have a copy of the original report somewhere too - the EMC guys asked it be taken down, but, well.
From memory it has the pics of the paper traces. The report above has those same graphs, but they've traced them into what looks like bitmap form. They were on paper.

What makes me despair is that the guy who's taken over the funding for polywell with the Navy seems to be "it's all plasma" guy. Also, I've spoken to the only guy in australia who has supervised a polywell experiment, and he's a plasma guy too. That experiment, btw, didn't show much potential well confinement, but they were using the supervisor's pet electron gun design, a can in a plasma. The plasma pressure in the polywell was too high - had to be for the can to make a "good" gun, an the current it produced was WAAY too low to drive the wiffle ball effect - they might have gotten closer by running their coils much more gently. Too hard and the cusp field makes an efficient pump for removing charged particles from its innards.

(Which BTW Bussard used - by firing off a magnetron just before each run, so that the neutrals within the wiffle ball were ionised by the electron resonance on the magnetic fields with the right intensity for it - thus improving the vacuum right when it needed to be empty.)

My biggest question mark with the polywell, is what effect the rate of change of the magnetic field itself had on the wiffleball effect. I suspect a fast rate of change of the field itself would have an enhancement factor on the force perceived on the electrons crashing into the field gradient - it certainly does on electrons in a wire.

There would still be such an effect even steady state, but the force on the electrons would be lower, due to it needing to be entirely by them feeling the "effective" rate of change of the magnetic field as they penetrate the extent of the wiffle ball and run into that suddenly intensifying tangential magnetic density.

So perhaps a smaller polywell could be made - but it would require pulse mode of even shorter duration. Easier then to make an oversize machine, and start out with very light large radius coils, with low peak intensity field (but still fast dB/dt). The geometry of the fast electron escape through the cusp points is size invariant in the beta=1 conditions, so bigger with fewer holes = easier to keep electrons within, and losses should go down if the electron pressure is lower owing to the larger radius too.

All rolls back to making a sufficiently large vacuum chamber. Need only be on the order of 10m across. Nothing too big, but annoyingly large enough that the pump-age needed becomes prohibitively expensive if bought off the shelf

For the polywell, higher charge per mass means the magnetic field does more. Miniaturising it to an "easy" size to play with would require electrons with proportionally *less* mass per charge. May well get a feel for the physics, but a working demo it'll never make. You're probably right though, I should just start building something for my own experience.

I have a friend of a friend who has his own property "out bush" as we say. He also has a couple of mass spectrometers and recently an electron beam microscope. I'm planning to go out with my connecting friend sometime later this year, and have a see whether he has space for say a large steel geodesic box... (Am intending to hold a rough vacuum in the strong steel outer shell, then have the high vacuum tank constructed from shim thickness SS sheet, so that it can be tig'd and held up without much effort. High vacuum pumps goes between. Power supplies and equipment get to go in the rough vacuum, but in their own boxes with external power and cooling water piped through. )

Anyway, must get to bed eventually, and I have another wall of text nearby here that actually didn't make this edit

Mostly on those damn sigma-delta adcs.

[chrismb]

oh boy!.. Match Doug's word count and I'm gonna be doing a lot of scanning and missing of detail!!

they're dead wrong and I have the measurements to prove that, in spades.
I've measured recirculation - it's nil, not even a little ringing on a ghz scope trace. There is not a neutral plasma or mainly nuclei - it's electrons in bulk, about 10x the number of positive ions.
The fusion isn't even mainly happening at the "focus" either

Regret to say, Doug, that I completely agree! I hope I've not just upset Frank over on fusor.net, picking on his arguments about recirculation on this very matter. Just that there's not enough info from 'regular' fusors to fully appreciate what is going on in them. Study a regular fusor and you, effectively, end up with one data point. You need different experiments to comprehend the inter-relationship.

Sorry, polywell gets a big thumbs down from me, for several reasons we can go into in another thread. The paltry neutron counts claimed could be easily explained by fast neutrals into the walls - which is pretty much how I have concluded fusors work (and is consistent with Doug's approach and Ti shell-lining).

The other thing to bear in mind is that if an IEC device - polywell, fusor whatever, but something that isn't thermalised - gets anywhere near useful power outputs, then this is incredibly problematic because you have to rid the chamber of all the fusion products because otherwise they will cause the contents to thermalise. However, pumping 'real power output' exhaust products requires pumps of billions of litres/sec. It's all quite impossible, I'm afraid.

..Looking forward to the debates on these points in their own threads. For now; welcome, Remy.