Capacitor bank trigger challenge

[Lerner]

Thanks for all the thoughts. We do need a fast rise—we want a turn-on ideally within 10 ns and definitely all firing within 100 ns. Each switch carries up to 250 kA with about a 1.5 to 1.8 microsecond rise time.
These are trigatron switches. Very simple design—described in the latest update from me. Here:
https://focusfusion.org/index.php/site/article/progress_in_solving_switching_problems/
We’ll have more to report on in a couple of weeks when most of the switches have the copper inserts.

[belbear]

Lerner wrote: We’ll have more to report on in a couple of weeks when most of the switches have the copper inserts.

Shouldn’t you go for tungsten coatings for the main electrodes? Copper is not THAT much more resistant to powerful arcing than the original aluminum, so it will only postpone problems. I fear cost for tungsten will be a major issue however.

[Pete Keech]

I would think Tungsten cost should be low enough, if the need is there, given the overall low materials cost of this device. You’re really only looking at thin Tungsten coatings, meaning hardly any material is used, so mass production will drive any cost down with time. My concern with tungsten is its potential to oxidize, as re-reduction is energetically costly. Reduction can also lead to big morphological issues, as this reduction of WO3 paper indicates…
http://www.springerlink.com/content/97587782t5p7n216/fulltext.pdf
As I mentioned in the other section comments,
https://focusfusion.org/index.php/site/article/progress_in_solving_switching_problems/P10/
I would think (conductive) oxide coated components would be more rugged, lower cost, and these are already being developed for many high temperature applications… (i.e. copper-ceria, lanthanum strontium manganite, doped perovskites, etc)…

[Henning]

Pete, that’s all just and well for the serial production. But here we just have an experimental device.

You’re an electrochemical engineer (or something like that), but LPP are all physicists. They can screw things together and set up measuring devices, but I doubt anyone can weld. Wolfgang the mechanic down the road does the milling and turning. (Isn’t there a better word in English? Something like “fräsen” and “drechseln” in German?) But I doubt he knows much about any kind of electrolysis. Probably just the basics, how you get a nice goldenly finish (I might be wrong here).

I would like to ask you for a favor: Could you maybe instruct Wolfgang what kind of finish is appropriate, and how to achieve it? Or maybe even do the electrolysis in your company’s workshop as a pet project (side project – or whatever it’s called), or as an official order by LPP?

I feel like you’re having heaps of great ideas, but Eric seems to be silent because he doesn’t have the means of bringing those ideas into reality (or maybe he’s just on holiday now). That’s why I’m asking you to step forward with a plan and a commitment to actually do the coating (I hope I don’t ask for too much). So the next time they’re unscrewing the switches and week worth of waiting they get a 50% performance increased switch.

If you’re able to do that, also contact Eric directly by e-mail through the forums, and ask him whether that’s actually desirable. I’m not writing on behalf of LPP, so ask them directly.

Cheers (and a great thank you),

Henning

[Pete Keech]

Henning – thanks for the compliment about my ideas. As I mentioned somewhere else on the site, I think many of my ideas are more suitable for scaling up / later generations. That’s because my knowledge is in material properties/preparations, rather than physics. I would also suggest that the last thing the project needs is to get bogged down making any component perfectly instead of proceeding through with the demonstration of the proof of concept (with possibly less-than-perfect components). At the point of net energy, there will hopefully be additional funding and interest in the project (and appropriate expansion). I have no objection to getting involved (now or later) with the project (I’m sure it would be fun), but it may be premature. (Does posting in the forums count as involved? I might already be involved if it does…) If there’s a true component failure (as opposed to sub-optimal performance), then we can address a specific issue in some way, if my help is desired… It is slightly complicated by the fact that I’m in Canada – an international border and a few hours away, but we can work that out, too.

Some general advice on a (smooth) finish on your materials & how to acheive it for this phase of testing if the materials are breaking down too much. Amusingly, I’d go more expensive (rather than less as I suggest above or in other places for the long term project goals). This assumes that what physicists do expensively, material scientists can do less expensively (given a proven system as motivation). I would coat components with Pt. Pt has a higher MP (1772 vs. 1083), forms a thinner oxide than Cu, is excellent for electrical components, etc, and has pretty much all improved properties you would want for Cu applications (at only 10,000X the cost!). The similarity of Pt and Cu (they are mined together) should make coatings relatively easy to acheive on Cu parts, as long as they have a reasonable finish. If not, then, that’s another issue (we can address that too, though). Often, getting films to stick is the difficult part.

One of the easiest ways, for mostly flat/smooth components is via sputter deposition, which naturally makes a pretty smooth surface (e.g. Pt will preferentally make a densly packed, predominantly (111) film). This avoids that messy complication of adding a chemist, and can be done quickly. Coatings can likely be verified with a simple microscope, or better yet an SEM (backscattered mode is best), and they can be assessed periodically as time proceeds. You can have sputtering done at any microscopy facility in between assemblies, and can take any spare Cu part to be assessed/optimized prior to trying it with your precision parts. You can also anneal the part, post sputtering to see if that improves/changes the coating.

If you have more tortuous surfaces, then sputtering may not work as well, but there are ways to address that, too. That’s where you get into the aqueous techniques (dip coating, electrodeposition, etc). If that’s the case, we can discuss this further, as we can about some of the more “advanced” coatings (oxides etc).

[Aeronaut]

Pete,

You’re already implicated by showing solid evidence of visualizing the machine and mass producing it. The more of our members who do that, the better off we all are.

I’d design all parts for maximum reliability, with a rapid drop-off in performance when servicing is required rather than “low price”, since it will cost thousands (at least) even for routine servicing, which would make a $5 savings meaningless. I’d also design for the likelihood that all 12 switches would need replacing within an hour or so of each other (run time). Replacing all 12 as a set could also be written into the regulatory requirements to keep FF from getting a poor reliability rap from those who try to run it too cheaply.

I doubt the electrodes would be a problem other than testing enough berylium-copper alloys and getting the dimensions worked out . An outfit like Brush-Wellman (I did some floors in their plant) could cast electrodes by the thousands, including threads, using the lost foam technique.

If I had your materials properties background, I’d be focusing on the capacitors and X-ray converter construction. Smaller, lighter cap banks and shielding as well as higher onion yields are going to fuel an ongoing manufacturing duel about who can make the smallest, lightest FF in a given power range.

[vansig]

Rezwan wrote: One of the challenges of the dpf seems to be getting the charges from the capacitor banks synchronized to within nanoseconds. This is an engineering challenge, and its solution will have an impact on the ability of the dpf to coordinate the charge that is to lead to net energy.

if charges are to be synchronized to within nanoseconds, then really we’re looking at carrying capacity of the wires in terms of high frequencies. to get the sharp-enough rise times needed to even measure a discrepancy, skin effects are therefore significant.

skin penetration depth in Cu @0.1ns is ~.65 micrometre. if you want current to rise from 0 to 2.3 MA in 0.1ns, then this constrains the diameter of your conductors to sizes much larger than i see in photos.

a 13.5 cm diameter copper pipe will rise to full current in 10 ns. is this good enough?

[vansig]

vansig wrote:

a 13.5 cm diameter copper pipe will rise to full current in 10 ns. is this good enough?

if we model as an RLC series circuit, then based on recently-measured 1.8 microsecond rise time, it seems like circuit inductance is a factor ~3000 greater than where it needs to be. what does this translate to? larger diameter conductor? coaxial shield? change in overall geometry?

Here is a hard line cable, used in broadcasting. what kind of feed wires are used in FF1?
http://en.wikipedia.org/wiki/File:Coaxialcableoneandfifthofan.jpg

[annodomini2]

Speaking of materials, Iridium?

[vansig]

vansig wrote:
what kind of feed wires are used in FF1?

okay, after viewing the wiring setup,
https://focusfusion.org/index.php/gallery/image_med/85/
i’m beginning to wrap my head around this. all the plates are arranged so that the electricity approaches the platform on a fairly flat, conductive surface, from outside to inside, which will tend to cancel induced magnetic fields, which is good. but what about parallel capacitance?

[achataignier]

Hi, just to know: why not use several Thyratrons in parallel, instead of this Spark-gap switch ?
(or even a crossatron, but newer technology will likely cause more problems)

Thyratrons and spark-gap switches are apparently close together technically speaking (both use a ionised gas to trigger), but Thyratrons seem reliable, at least much more reliable than spark-gap switches given your descriptions. They have been used since many years in the industry with success.

Sorry if the answer is obvious for you, I am not a plasma specialist;-)

[jamesr]

I have been thinking that trigger system looks quite complicated. Can someone explain why there is a separate trigger on each capacitor? If they are all charged in parallel to the same voltage, can’t you just have one spark gap in the middle?

If you do need separate triggers why not use some of the techniques used in Marx generators. Such as all the gaps should be able to see each other, so the UV light from one spark can help trigger the others. Then the time difference between them is down to how close together they are. The other step that is apparently used, is to dope the electrodes with a radioactive isotope, to give a consistent breakdown point.

James

[Aeronaut]

The switches have to carry a huge amount of power and current, fortunately for only a millionth of a second every thousandth of a second in ballpark figures. This low duty cycle helps a great deal with cooling and power lost as heat, but is almost science fiction when you shop for ways to switch that much power that fast and precisely without paying too much in inductance.

My first thought is always use a bank of stud-mounted SCR silicon controlled rectifiers, which I’m sure would cost less money, at the cost of way too much inductance.

My proposed solution is to turn the physics, math, and engineering departments of around 2,500+ universities loose on switch technology, capacitor technology, and the onion’s production systems. Somewhere in that maelstrom is at least one pocket of people who will find a way.

I’d also bet on that group to solve several of the key production engineering challenges. This is how we speed up time to market, or, in the worst case, make sure the engineering phase takes no more than 3 years.

[annodomini2]

Found this in a little search:

http://www.abb.ch/product/ap/db0003db004291/c12573e7003304adc1256efc004d2922.aspx?country=US

Claim to have High Voltage/High Current solid state pulse power supplies, doesn’t specify details just contact information, but may be an avenue of investigation

[Aeronaut]

annodomini2 wrote: Found this in a little search:

http://www.abb.ch/product/ap/db0003db004291/c12573e7003304adc1256efc004d2922.aspx?country=US

Claim to have High Voltage/High Current solid state pulse power supplies, doesn’t specify details just contact information, but may be an avenue of investigation

I saw that page again yesterday, along with the solid state switch family they’re describing. Although the voltage and current limits appear to be way to low for a single package switching solution, the power overload delimiting table wasn’t mentioned. Since these are obviously intended for high power pulsed supplies, it could be worth contacting ABB about.

[Author]

Posts