Capacitor bank trigger challenge

[Breakable]

Tnx for the update. Some random ideas:
Is it not possible to test the switches on low power and record the data somehow? Even nicer it would be to be able to record data with high power for run time diagnostics. Maybe few mega ohm resistors would be able to shield the voltage/current from (ultra-fast) sensors?
Also I wonder why you need separate switches? Is it just the power requirement or you need to control the firing? It might be simpler to have one switch and adjust the firing sequence using wire length, of course you need diagnostics and I am not sure if you can get any switch in that power range…
Do we have any real electrical engineers here?

[mchargue]

Timing on switches… How about this.

If you’re using spark-gaps, or something similar, is it possible to hold them all slightly below their spark threshold? Then, with them all in this state, you can trigger them all using a laser, or some other excitatory means, to produce a conductive plasma.

Unless you’re using an ignitron, or some other enclosed gap, flashing a plasma near the gap should be enough to close it. If the laser were then fired over all gaps, that should have all the switches closing at the same time – less light travel time. Whatever delay was engendered by the distance from the laser could be compensated for by shortening output leads based on their distance from the laser light source.

You know, there may be some good ideas out here, but you’ve given almost no information on what system you have in place now. It’s hard to just keep punching in the dark on this.

Got time to describe the system you’re using now?

Pat

[Axil]

Chech out tis reference, there are some z-pince diagrams discribed.

http://arxiv.org/pdf/0809.2071

[JimmyT]

So the switches are dependent upon suddenly creating a ion bridge (plasma bridge) between the capacitors and the electrode? And this ionization is dependent on the potential between the two plates? And when the first switch or two is triggered this potential difference becomes less, so subsequent switches don’t fire? Is it possibly that simple?

It would be a cast iron B**ch to trouble shoot if you are assuming that there is something wrong with an individual capacitor or switch.

[Breakable]

In any case sensor output for diagnostics would be a good idea, I just wonder how difficult it can be to measure such high voltage/amperage.
Voltage might less of a problem than amperage.
Maybe measuring the temperature of a wire could say something?

[mchargue]

Axil wrote: Chech out tis reference, there are some z-pince diagrams discribed.

http://arxiv.org/pdf/0809.2071

I’m not certain about this, as there’s not a lot of information on the capacitor system, switches, or its connection to the FF reactor, but I don’t think that a super-Marx generator is what they’re using. I don’t think that timing would be such an issue if this were being used.

The Marx device will step up a low voltage to produce a high voltage spark by charging capacitors at a low voltage in parallel, and discharging them in series through many spark gaps. More likely, I’d say it’s the opposite: charging capacitors with a high voltage in series, and discharging them in parallel for a high current.

The topologies exist to do this using diodes, but I’m not sure you can get diodes with sufficiently high power ratings. (though they may be available) Without diodes, you’d need a lot of spark gaps, and timing would be critical for closing each. (hence, the challenge?)

See,
http://koasas.kaist.ac.kr/bitstream/10203/6494/1/%5B2008%5DPower-Efficient Series-Charge Parallel-Discharge Charge Pump Circuit for LED Drive.pdf

To dispense with the requirement for high-current diodes, you’d string all the capacitors together, nose to derriere using a (low-current, low Vbr) diode. Then you’d tie the series connected lot across a high voltage potential to charge it. Each capacitor’s negative & positive terminal is then connected with its own spark-gap to the FF reactor’s negative & positive terminal. (two per cap)

To fire the bank, the top & bottom and all spark gaps would be ‘closed’ together to short the entire bank of capacitors to the FF device in parallel. Bingo! High current from a high voltage source.

This topology might work better with the voltages expected voltages generated by the ‘reverse linear accelerator’ that taps the power from the FF reactor, but that’s probably not a prime consideration. This would dispense with (possibly non-existent) high power diodes, but would require the use of a pair of spark gaps for each capacitor used. A tractable challenge, I’d say, and a trade off required by what’s available.

[mchargue]

One other idea…

Does the shape of the current pulse matter as much as the amplitude? If not, then maybe to get the amplitude without having to worry about the timing so much, just add some more capacitors.

If some come in late, or early, it may not be such an issue. So long as a majority come in together, and produce a sufficient amplitude, you’re golden. This would be especially useful if the timing of the bank varies from shot-to-shot. You’re just looking for a majority to be on time so that you can ignore the outliers.

Failing that, is there some way to get the capacitors to time their own firing? Like in a super-Marx generator, where earlier stage spark-gaps load up on later stage spark gaps as the potential across later stage spark-gaps keeps increasing.

Can the capacitor banks be coupled somehow to reinforce each others’ firing?

[Henning]

Actually timing is crucial. All capacitors have to shoot at the same time. If one capacitor is early, the sheath already runs away before the others set in. Or when it’s late that capacitor just puffs off its energy.

Main thing here is acceleration, so the final destination is reached at top speed.

[Dr_Barnowl]

Stupid question :

Why won’t it work to just connect all the caps in parallel and put a big switch on that single bus?

I know it’s a stupid question because I’m not an EE.

[Henning]

Because there is no switch that big and that fast. One solution was to use a diamond laser switch (diamond is a non-conductor, but if an ultra violet light shines on it, it conducts pretty well). But as I read here on the forums, there isn’t even one of those switches that big (yet). So they used the approach with multiple switches, and not diamond laser switches.

[Axil]

Is this fast enough?

This paper describes the design and implementation of a driver that allows an IGBT module rated at 800A/3300V to switch up to 3000A at 2200V in 3μS with a rate of current rise of more than 10000A/μS,

http://www.slac.stanford.edu/cgi-wrap/getdoc/slac-pub-12591.pdf

http://home.earthlink.net/~jimlux/hv/pasley1.htm

If you want to switch 50 kilo Amperes with a sub 20 nanosecond commutation time at 20kV you are going to be in trouble if you are relying on semiconductor technology. However there is an alternative class of devices that have been around long before the humble transistor came on the scene. You might think that vacuum tubes and similar are a thing of the past. But for problems of this magnitude they are the only things on the market that will do the job.

This application note describes the use of nanosecond SCR switch (thyristor) to enable circuit designers to upgrade high-current, high-voltage modulators and pulse generator circuitry.

http://www.eetindia.com/ARTICLES/2001SEP/2001SEP20_AMD_AN1.PDF?SOURCES=DOWNLOAD

All these solutions are in the lightning bolt range, you should find here something you like. If not, say why.

[Pete Keech]

This discussion has lead me to wonder about a possible design simplification, maybe on a 2nd or 3rd generation system, that could potentially boost energy budgets.

Once the process was ignited from the large capacitor bank, would it be feasible to input the appropriately tuned electrical pulse output (i.e. after conversion of emitted He2+/x-rays to electricity) directly back into the plasma to induce successive shots? By tuned, I mean regulate the input electrical energy to be the appropriate value by “bleeding off” the additional gained energy into your power grid, and simply redirecting the appropriate output back into the input of this pulsed system. The alpha emission approaches unity energy on its own according to this article:
http://arxiv.org/ftp/arxiv/papers/0710/0710.3149.pdf

so a slight boost from x-ray energy directly (or a much smaller capacitor bank charged from previous events & timed to the output) seems like a reasonable way to eliminate “electrical wear & tear” on the system. If overall timing is an issue (too fast), then a series of units could reduce it to manageble level (10 machines in series slow timing by factor of 10 to give 30 MW instead of 3 MW output, 100 slow timing by factor of 100 to give 300 MW, and these are still in realm of conventional power generation). This configuration would allow you to use a single large capacitor bank to initiate any of the 10 (or 100) units, and then small capacitor banks on each one to “tune” their inputs. Minor engineering would enable failed units to be dropped from the circuit, or even allow power generation to occur in the event of servicing (e.g. 110 units, 10 at rest/being serviced at any point, generating 300 MW continuously).

In terms of energy budget, this strategy may also allow a much stronger/more efficient (Blake-inspired) magnet to surround the entire system. MRI or NMR are in the order of GG (or higher).
http://www.umbc.edu/nmr/safety.htm
so that technology could easily be adapted to this sort of set up.

[mchargue]

Here’s something… Using a DPF as a switch for an inductive (as opposed to capacitive) energy storage system. An interesting read, and you folks might be able to better understand how this might apply to your own work.

http://www.patentstorm.us/patents/4406952/description.html

BTW, I enjoyed reading up on the progress w.r.t. the spark-gap switching system. Glad you folks have a handle on it. I look forward to more technical details on this, and other aspects of the FF system.

Pat

[Dr_Barnowl]

Are these spark-gap switches a scaling up of a krytron?

[mchargue]

Dr_Barnowl wrote: Are these spark-gap switches a scaling up of a krytron?

Sure sounds like something close to it. I think, though, that they dispense with the grid, and initiate a conductive plasma by jumping a high-voltage spark across the gap. After the conductive plasma is started, it persists until the capacitors finish discharging through the plasma – sustaining it until the current drops off.

Pat

[Author]

Posts