Drift tube accelerator project

Various things to make charged particles fly

Drift tube accelerator project

Postby Doug Coulter » Tue Oct 11, 2011 4:18 pm

I am temporarily moving away from fusors and into beam devices so I can more easily learn some things in a simpler system (for the particles, the gear is more complex). Thus, I've been acquiring knowledge and parts to build a beam device, with target energies per beam in the 100kv region. Being that voltages much over 50kv are a real pain to handle in air, and that I want to be able to achieve decent energy spread and focus along with good bunching, I selected the not-very-versatile drift tube design for this project. This will let me get there in decent size with the still formidable 12kv or so RF peak voltage for D ions, and with some changes, for the other ions of interest. The sharp will note that 100kv is right around the peak cross section for DT, in beam on target applications. The plan might be to make two of these and do beam on beam, but I think it wise to make one, solve it's problems and unexpected issues first, then add another, which might be optimized for some different charge/mass ratio and output energy anyway. I wrote some software to help calculate the tube (or really, gap to gap) lengths, which is here. Fairly rough stuff, but it's not going to be used too many times - you get the answer, you're done with that tool. So I didn't bother with minor things there, like a nice GUI, and I'll need to subtract gap lengths manually, oh my.

Since this isn't new science, nor is it really new tech, I'll let Haliday say most of the words. Note the page on focusing has the only actual error I've detected in the whole book - a particle spends LESS time in the second half of the accelerating gap, not more (and see, the sentence makes more sense that way).
TDT1.gif
Haliday on drift tube accelerators #1

TDT2.gif

TDT3.gif


Per Terman, only the first gap between the ion source, and whatever goes on inside the ion source itself really affect total focus. With tubular elements, you only get a measureably small focal length when you're adding a lot more energy to a particle - the velocity difference between one end and the other end of the lens gap has to be large to get a lot of focusing action. One can improve this, as Haliday notes (as does Terman, but with math and equations) by using a smaller aperture on the inlet side of the gap, going downstream, which I plan to do. This also cleans up the beam, as things that miss the aperture at least don't get down the tube for more energy to be wasted on accelerating them further. Thus, I should be able to use pretty short inter tube gaps once past that first one, which makes things smaller, and simpler to work with, but I'll have to be able to adjust that first gap, and whatever extractor/lens I put in the ion source. I plan to use the microwave one we've kind of pioneered here, it works great for things like this, and works down to way low pressures like you want in a beam accelerator, no differential pumping should be needed at all.

But that first gap, going from ground (the ion source output tube will be ground, it will have a 20kv "pusher" at positive polarity), to varying RF voltage is certainly going to be interesting. With 20kv or so DC on the ions at that point, seeing a voltage from -11kv to + 11kv means the speed and focus at that point are going to be crazy, and time-varying. This should actually help with initial ion bunching, but finding a happy place for that lens design is going to have to be cut and try -- too hard to model, I bet even SIMION would have issues trying. But we have an existence-proof - we know these are made and that they work fine, so we'll just follow the footsteps of the giants we stand on the shoulders of, like usual. The fact that there are a good number of downstream gaps for both time and space bunching should give us a nice beam quality. In the mechanical design, based on the cross I think I've pictured elsewhere here (I'll link from here when I find it again), we'll be set to do beam on target, or beam on beam, with decent diagnostic capability -- I've been thinking on this one awhile. Since the most off ground voltage near the ion source is a mere 20kv for "pushing" the ions out, I can use that unmodified, the insulation of the ion source beam tube will take that easily, and nothing else has to get off DC ground at all. I note that the tubes should be at least 3 diameters long for the middle to not see too much field from the ends, and with a 1" beam tube, I satisfy that pretty easily, I can go to almost 3/4" diameter drift tubes and fit inside that easily -- and 1/2" will fit with a lot of quartz "rigging and jigging" as is used in say, a Tektronix CRT, where the same issues of alignment apply, so perhaps we copy how they built those. Now that I know rough dimensions (from the software) I can put one of these together and put it on a capacity tester. A big goal will be to keep the capacity to ground and from phase to phase of the RF absolutely minimal, as the circulating currents required in the tank circuit will be fierce no matter what (and those make losses). I figure perhaps 100 watts max into the beam, but this could easily require several hundred watts losses in tank circuits due to that circulating current and parasitic resistance in the tank coil....fun! I'm planning to use a 3-500z power triode for the RF output amplifier, gain modulated (volts are a picky part of tuning) to support this, and use the 13.56 mhz ISM band for the RF, since it's legal to accidentally radiate some in that band without a license. Onward and upward, probably in fits and starts, but here we really have a start - I have all the pieces I didn't plan to make anyway, and a rough design now.
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.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Drift tube accelerator project

Postby Doug Coulter » Tue Oct 11, 2011 5:30 pm

Like I said, the drift tube accelerator is a little picky and is a mass spectrometer. Here's the results of a couple of runs that show that with only changes in both DC and RF voltages, you can indeed accellerate different ions...You pretty much just scale the volts by the mass/charge ratio of the two ions, then can use the same hardware (if it will DO the different voltages).

Code: Select all
doug@main:~$ drift
Choose an ion to accelerate, one of:H:T:D:3He ->d
you chose D, 1870945442 eV mass
Injection voltage? [20000] ->
Injection volts = 20000
RF Frequency? [13560000] ->
Frequency = 13560000
RF peak volts? [11500] ->
peak volts = 11500
Number of RF driven tubes? (should be even #) [4] ->
RF driven tubes = 4

Net beam energy = 112000 eV


rf tube 1, volts in tube: 31500, tube length 6.41356820951236 cm
rf tube 2, volts in tube: 54500, tube length 8.43611890040296 cm
rf tube 3, volts in tube: 77500, tube length 10.0599377247279 cm
rf tube 4, volts in tube: 100500, tube length 11.4558540471269 cm
grounded tube, beam volts: 112000, tube length 12.0935401934631 cm
total length (minus last) 36.3654788817701 cm, 14.3171176699882 inches
doug@main:~$


doug@main:~$ drift
Choose an ion to accelerate, one of:H:T:D:3He ->t
you chose T, 2808818208 eV mass
Injection voltage? [20000] ->30025.65595924
Injection volts = 30025.65595924
RF Frequency? [13560000] ->
Frequency = 13560000
RF peak volts? [11500] ->17264.752176563
peak volts = 17264.752176563
Number of RF driven tubes? (should be even #) [4] ->
RF driven tubes = 4

Net beam energy = 168143.673371744 eV


rf tube 1, volts in tube: 47290.408135803, tube length 6.41356820951238 cm
rf tube 2, volts in tube: 81819.912488929, tube length 8.43611890040297 cm
rf tube 3, volts in tube: 116349.416842055, tube length 10.059937724728 cm
rf tube 4, volts in tube: 150878.921195181, tube length 11.4558540471269 cm
grounded tube, beam volts: 168143.673371744, tube length 12.0935401934631 cm
total length (minus last) 36.3654788817702 cm, 14.3171176699883 inches
doug@main:~$



Yeah, I used the ratios calculated from the (hopefully corrected correctly) mass numbers inside the code. Those require quite an increase to go from D to T, roughly 3/2 as you'd expect. I'm using this program to investigate that sort of thing, an this might militate to using higher frequency, shorter tubes, and lower voltages to keep things reasonable -- I'd love to be able to do lithium for example, and I don't need the higher beam energy. For DT you only want ~110kev total. But for DD beam on target, you want at least 125kv, so perhaps the real advantage is that these can easily be made and inserted in a simple quartz beam tube - substituting one should be pretty easy. I'm still exploring the tradeoff space here. There's another ISM legal band at 2x this frequency, and one at half, and crystals and so on are common for these frequencies. They laid out the allocation this way as harmonics are often generated in cheap bulk industrial RF sources, so if you radiate some 2nd harmonic, you're still legal and not interfering with anyone allowed to complain about it. If I was running a few or tens of watts, I wouldn't worry much, but at KW levels, even a small antenna will let some significant power out into the world, especially at these voltages... could be going to more drift tubes has other advantages -- better bunching for one, but some disadvantages, more capacity being obviously one.
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.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Drift tube accelerator project - Tektronix CRT technolog

Postby Doug Coulter » Thu Oct 13, 2011 6:24 pm

Just thought I'd get this link up here so it wouldn't get lost again, and thanks BillF for finding it again for me (sadly I forgot who first sent it. Jerry?). The techniques used to maintain alignment on an old tektronix electron gun look perfect for this project, and I'll probably use them. I didn't catch them the first time I saw this movie, but those guys were pretty darn slick with tricks and neat jigs.
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.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Drift tube accelerator project

Postby chrismb » Fri Oct 14, 2011 2:45 am

Doug Coulter wrote:Please read the Haliday I scanned at the link above, it explains how bunching works (and why it won't with a square wave) as well as anything I've seen so far. And it DOES work with sine waves, or it wouldn't work in cyclotrons, linacs, CERN and so on - except, it does work in all those places - the design documents are in our library too, and their results speak for themselves. You need a time varying field that accelerates the laggards more, and the ones that are ahead less, to keep the stuff bunched. A vertical risetime does not satisfy the conditions. Ion traps use sines too...


You can call it any way you like, Doug, it's your experiment and forum.

But Haliday doesn't discuss the point I'm getting at in the pages you've posted, so there's nothing to read about it. What Haliday doesn't discuss on that page 291, fig 5, is that as any particle tend towards phase a, once they get there any small temporal perturbations that put it on the b side of a mean it is lost because there is never then enough e-field to get the particles back up to the speed of the rest of the bunch. a is both an attractor and the minimum viable location in the phase, which is a recipe for a 'loss cone' for particles to exit focussing if they happen to undergo any down-scattering that push them further back than a in phase.

He put a 'b' there but didn't mention it again in the context of phase focussing. Particles will bunch around the 'phase attractor space (a' to V<hat>)', and in doing so some will be on a tail of distribution on the 'wrong side' of a.

So, although he discusses pulling forward the 'laggards' in the a' to V<hat> region because the phase attractor is a whilst the 'operator' field is set ahead of a, he makes no indication of what would reduce losses for those laggards that get further behind than a. Those are lost, but this is a situation mitigated by a square wave that (in theory) has the same acceleration vector at all points in the positive part of the phase, so the 'loss cone' is reduced.

I did my calcs for my device and the losses using sine were way to high by this mechanism. A square wave tends, more so, to push the particles on the a' side of a because there is no let-up in the accelerating field. In other words, set the square wave potential to V<hat>, slightly higher than the energy you want the particles to be at, and though the phase attractor is a, the phase grouping tends towards a', which is a more stable configuration. By having the e-field slightly higher than the target/design velocity for the applied e-field period, all the particles will try to push towards the front of the square wave. This is then 'stable' because you end up with a distriution of particles rushing ahead of the wave, which can then be swept up if they slow down a bit, whereas with sine they will bunch at the design phase and the distribution of particles will mean some are ahead, but some are behind too and never catch up again. Well, this is how it is for my device, and I don't immediately see why it would differ much for most other periodic-gap accelerators.

I'm not saying sine doesn't work, I'm just saying there is a loss cone of particles out the back of the phase that can be mitigated by more-squared waves. And in practice, bear in mind long ago the best they could do at MHz was sine, and even if you do generate a pure sine, by the time you put an accelerator load on it then the power signal will undergo compression and end up more square-like anyway. What is v.v. bad is if you end up with a very spikey power signal, for example as a result of inductive effects, which is the 'opposite' of moving towards a more square-wave. If you end up with your sine becoming spikey then the phase attractor a to V<hat>, will narrow so that you end up with a much larger tail of the particle distribution ending up on the 'too laggardly' side of a never to be pulled back into the bunch.
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm

Re: Drift tube accelerator project

Postby Doug Coulter » Fri Oct 14, 2011 9:24 am

In a (normal?) particle accelerator, there's no way after the first stage that things "get behind" by very much - only coulomb-spreading of the bunch, which has minimal time to operate stage to stage, and is "fixed up" at every gap by fields far larger than the self-fields of the ions (unlike in a fusor where the applied and self fields are comparable due to high current density/voltage ratio). There are effectively "no" collisions. The mean free path is many times the path through the accelerator. All the particles that don't get into the bunch on the first stage are lost, before you invest (waste) any more energy on them (more than 99% in most designs). They might get another chance on the next cycle, but you really don't care - you only wasted roughly the ionization cost plus one stage's V worth at most trying to gather them up. Most of them hit the wall and become neutrals. But there is so little of anything whatever in the there that inter-particle collisions are nil in these devices (until the target, of course).

Going down the pipe the rest of the way they oscillate around the ideal phase a little, as Haliday says. Not all devices are run in thick gas to get collisions (which is the only way for things to "get behind" - there's no other friction) - you're viewing the entire world through the lens of your own (proposed) device that runs under vastly different conditions than, well, everything else there is that is in use and working so far. What I'm attempting here is simply to build an accelerator to use the beam(s) for experiments, very conventional, actually. The scheme presented here is known to work quite well, though few or none have tried it on such a small scale and at such low energies (mere 100's kv). To the extent I push the state of this art at all, it will be with "high" beam currents - in the 1 ma range, vs the few micro-amps most of these things do (at best). I'm hoping that combining the standard accelerator tech with a superior ion source will help me get there (or close enough). One feature of this not much discussed yet is that this ion source works well down to such low pressures no differential pumping is required to get that long mean free path in the rest of the apparatus, and puts out ion currents "to die for" vs most of what's described in the literature for sources that have any meaningful lifetime in use, and at far lower gas pressures (while fitting in a room). While an anode layer ion source makes more ion current, it only works in pressures a factor of 1000-10000 more than this, which would indeed lead to frequent particle collisions in the beam tube and things "getting out of phase" enough to be off a whole half cycle.

As to what would work best in your device, it can't even be modeled or guessed at until some specs are given for the mean free path for collisions between the moving ions and the neutrals, if you're still saying that's how it is going to work, vs the original way you proposed for two colliding beams of ions in it in that patent you keep forgetting to actually post up here and allow discussion on. Isn't that a little intellectually unfair? It would seem to the casual observer that if you've got a very short mean free path to cause that difficulty, you won't get particles up to enough speed to overcome the coulomb barrier at collision and create fusion when they do with a mere 500v or so per stage as it were. The lack of that information makes it look like you're trying to have it both ways at the same time, a sin of omission in your modeling?
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.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Drift tube accelerator project

Postby chrismb » Fri Oct 14, 2011 4:33 pm

Fair comments. Yes, I can see it could be important to actually want the particles in a problematic phase to drop out early, because, as you point out, I'm aiming for collisionality whereas a beam process wants to avoid that.
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 6:32 pm


Return to Accelerators

Who is online

Users browsing this forum: No registered users and 1 guest

cron