High-energy LC oscillatory discharge: what kind of switch?

[Majorana]

TL;DR Summary

EMP commercial and military generators produce the "classic" UNIPOLAR pulse, usually by discharging a Marx generator into a suitable antenna (often some kind of dipole). But my need is to produce an OSCILLATORY discharge into an inductor... I guess that having an oscillatory current is quite a problem for the switching element...

Hello everybody (after quite some time!)
I need to select the proper switch to be used in a high-energy, high-voltage LC circuit, intended to produce an oscillatory discharge through the L element. I started my research from the already existing equipment, as it would be a waste of time and resources to re-invent something that, most probably, is already in use somewhere.
Most of you probably know already the waveform of the "classic" NEMP:

This pulse is generally obtained by discharging a Marx generator (stack of series capacitors) into a suitable antenna, usually some kind of dipole. The pulse is clearly unipolar (like the pulse produced by a real nuclear device going off), there is no ringing of the output voltage, and the current flowing through the series circuit formed by capacitors-switch(es)-antenna, is unidirectional too.
Unidirectionality is a good thing when having to use switching elements like triggered spark gaps (aka "controlled spark gaps" or "3-electrode spark gaps") or hydrogen thyratrons, because all these switches cease to conduct and reset back to high-impedance state when the anode-cathode current drops below a certain level and the ionization extinguishes.
But in my case, I don't have to generate any EMP/NEMP. All I need is to discharge the capacitor(s) into an inductor, to obtain an oscillatory, damped discharge, like this:

So the circuit will be a very simple series of 3 elements: C [with its associated charger] + L + switch
(...I guess I don't need to draw it... )

The switch is the issue.

As far as I know, a triggered spark gap or a thyratron will turn off (ionization ceases) even before the current reaches zero: it is sufficient that the current drops below the minimum value necessary to keep the ionization going. For that reason I suspect that such a switch would brutally open at (actually before) the first zero-crossing of the current, so I would get quite a remarkable - and possibly destructive - voltage spike from the inductor that finds itself suddenly disconnected from the capacitor.

Please correct me if I am writing nonsense.

So the problem is: what kind of switch may I use in such an application?
I don't even consider krytrons because, apart from the well-known problems of buying them legally, I think that they feature the same issue of resetting to open when the current drops below a minimum level (and they are almost certainly too small to handle tens of Joules at 3000-5000 V for hundreds or thousands of times, for sure not a big issue when they have to set off an A-bomb... ). So no krytrons here.

What can I use???

Many thanks in advance to everybody!

[Post edited by a Mentor]

 

[tech99]

A spark gap will continue to conduct for many cycles. and will have a resistance in the order of 0.5 Ohm. This is how how high peak-power radio transmission operated at the beginning of the twentieth century.

 

[DaveE]

SCRs should do it. 2 back to back with sustained triggers, or one with an antiparallel diode. Triacs probably aren't big enough, IDK.

 

[berkeman]

handle tens of Joules at 3000-5000 V

Is that a typo? What L and C values are you using?

 

[tech99]

Can you say what peak power you are aiming to achieve. In his laboratory experiments, Hertz obtained in the order of many Megawatts.

 

[DaveE]

I suspect that such a switch would brutally open at (actually before) the first zero-crossing of the current, so I would get quite a remarkable - and possibly destructive - voltage spike from the inductor that finds itself suddenly disconnected from the capacitor.

Whatever switch you use, you will want snubbers or clamps for several reasons. Zero current switching is kind of a myth, it's really "almost zero" IRL.

 

[Baluncore]

If the main switch is integrated into the Marx generator, then the switch voltage will be lowered.

Consider having different technology switches in parallel. For example, a spark gap can be in parallel with a MOSFET. The MOSFET is arranged to conduct after the spark, then remain conducting during the following sinusoidal decay.

Most gas discharge lightning arrestors will stay on for some time after being triggered. For a sufficiently high frequency, they will not turn off.
What frequency of oscillation do you anticipate?

If you want a local magnetic RF pulse, there is a circular form of the Marx generator. It is a single turn loop antenna.
A NOVEL TYPE OF HIGH POWER PULSE TRANSMITTER
By K. LANDECKER* and K. S. IMRIE* 1960.
http://adsabs.harvard.edu/abs/1960AuJPh..13..638L

 

[Majorana]

A spark gap will continue to conduct for many cycles. and will have a resistance in the order of 0.5 Ohm. This is how how high peak-power radio transmission operated at the beginning of the twentieth century.

Yes, the ionization in the discharge gap takes a finite time to extinguish, so if the slew rate of the current is sufficiently fast, it can cross the "extinction zone" (0±"minimum keep current") fast enough so not to let the ionization extinguish. I'm afraid that it's a figure that must be indicated in the datasheet by the manufacturer of the spark gap itself...

 

[Majorana]

SCRs should do it. 2 back to back with sustained triggers, or one with an antiparallel diode. Triacs probably aren't big enough, IDK.

Uhm... I'm afraid I don't feel comfortable shorting (well, almost) a large capacitor charged to a few thousand volts through a semiconductor device. The multi-kiloampere pulse is still the realm of the sturdy discharge tubes. I think the SCRs should be well oversized to work reliably. Is there any model able to withstand at least 3,000 V? If not, two or more in series should be used, with voltage-equalizing resistors and more problems due to the different potentials of the various gate terminals. On the other hand, a single discharge tube does the job and is much less delicate than any semiconductor.

 

[Baluncore]

Uhm... I'm afraid I don't feel comfortable shorting (well, almost) a large capacitor charged to a few thousand volts through a semiconductor device.

You are not shorting the capacitor.
There is an inductor in series with the switch, that limits the current rise-time.
v = L ⋅ di/dt ;

 

[Majorana]

Is that a typo? What L and C values are you using?

Hi! No, it's not a typo. I'm still working on the figures, and it's not easy... among other things, almost certainly there will be a twinaxial line - to take into account (the losses) - between the inductor and the capacitor+switch. Voltage will sit somewhere between 2 and 5 kV (with less than 2 kV it would be difficult to use a controlled spark gap), stored energy will be not less than 50 Joules.

 

[Majorana]

Whatever switch you use, you will want snubbers or clamps for several reasons. Zero current switching is kind of a myth, it's really "almost zero" IRL.

Actually, I am not aiming at any zero current switching. I just need to discharge the capacitor into the inductor and let them oscillate until extinction. I mentioned the "zero current" thing only because discharge tubes tend to open spontaneously (even if I don't want...) whenever the current is about to cross zero. But in this application, for sure I don't want to interrupt anything, I would let it ring until energy is depleted.

 

[Majorana]

If the main switch is integrated into the Marx generator, then the switch voltage will be lowered.

Consider having different technology switches in parallel. For example, a spark gap can be in parallel with a MOSFET. The MOSFET is arranged to conduct after the spark, then remain conducting during the following sinusoidal decay.

Most gas discharge lightning arrestors will stay on for some time after being triggered. For a sufficiently high frequency, they will not turn off.
What frequency of oscillation do you anticipate?

If you want a local magnetic RF pulse, there is a circular form of the Marx generator. It is a single turn loop antenna.
A NOVEL TYPE OF HIGH POWER PULSE TRANSMITTER
By K. LANDECKER* and K. S. IMRIE* 1960.
http://adsabs.harvard.edu/abs/1960AuJPh..13..638L

I will not use a Marx generator. Those are used when you need really HIGH output voltages, beyond the capability of single capacitors. I don't need to go up to such stellar voltages, this is not an EMP generator, 5 kV should be enough for me. 5 kV capacitors are commercially available, no need to connect them in series in a Marx configuration. I am still working on the figures, but the oscillation frequency will be certainly below 100 kHz. Discharge lightning arrestors aren't triggered, i.e. they are uncontrolled 2-terminal tubes, generally in a ceramic case (I prefer the glass case because it allows checking the electrodes for erosion or damage).

 

[Majorana]

A NOVEL TYPE OF HIGH POWER PULSE TRANSMITTER
By K. LANDECKER* and K. S. IMRIE* 1960.
http://adsabs.harvard.edu/abs/1960AuJPh..13..638L

Will check this one right away, thank you SO much!

 

[Baluncore]

I will not use a Marx generator. Those are used when you need really HIGH output voltages, beyond the capability of single capacitors.

Marx generators have the advantage of lowering the switch voltage and the capacitor voltage. There are low voltage Marx generators that use CMOS logic, signal transistors, and produce 100 volt pulses.

The critical figure is the peak current you require to flow during the first pulse.
I wonder if a single capacitor will handle that voltage and ESR heat.
A handful of 500 V, pulse rated capacitors, designed for use in snubber networks, would meet specifications in a Marx generator, without the 3 to 5 kV requirement.

 

[Majorana]

SCRs should do it. 2 back to back with sustained triggers

You have just suggested me something. Not with SCRs but with 3-terminal ("triggered") spark gaps. A triggered spark gap is triggered much in the same fashion than a Xe flash lamp, i.e. applying to the trigger electrode a brief high-voltage, high-frequency burst that initiates ionization. That burst is usually obtained by discharging a small capacitor into the primary of a small step-up transformer. Okay: what if, instead of a brief high-voltage, high-frequency burst, I apply the same signal but sustained, 100-200 ms at most, in order to ensure that the spark gap stays continuously triggered during the oscillatory discharge? There would be no need to maintain a high slew rate of the main current in order not to allow the gap to de-ionise. Uhmmm... for sure I would need to hear the opinion of the spark gap manufacturer as that is definitely NOT an usual technique, but I think it would be worth a try...

 

[DaveE]

Yes, this is (was?) common way to start Xenon flashlamps for lasers. It can definitely work if you can get the E-field in there. The other common way is to spike a low power HV pulse across the flashlamp usually blocked from going backwards by an inductor or HV diode. In a similar way a ground plane near the lamp envelope will aid starting by increasing the E-field near the anode. I think all gas discharge based switches are basically triggered avalanches. That's why they're fast.

Fast SCR switches can be made with a series stack where one is triggered conventionally and the rest of the string avalanches. It's really old tech, and may be difficult for high power. But we did it to replace krytron switches in a laser Q-Switch driver many decades ago. GA301A comes to mind. The idea is similar, trigger an avalanche by switching part of the voltage.

 

[Majorana]

If you want a local magnetic RF pulse, there is a circular form of the Marx generator. It is a single turn loop antenna.
A NOVEL TYPE OF HIGH POWER PULSE TRANSMITTER
By K. LANDECKER* and K. S. IMRIE* 1960.
http://adsabs.harvard.edu/abs/1960AuJPh..13..638L

Got the PDF... brilliant, really brilliant! That Aussie was a master of his trade!

 

[Baluncore]

Got the PDF... brilliant, really brilliant! That Aussie was a master of his trade!

Take great care to follow the national regulations and avoid interference with other services.

You should be aware that building, and then operating such a device in the open, would rapidly attract unwanted attention from the regulatory authorities. It will sound alarms when it first appears, and will be logged on the national lightning detection network. You will have nowhere to hide.

 

[Majorana]

Take great care to follow the national regulations and avoid interference with other services.

You should be aware that building, and then operating such a device in the open, would rapidly attract unwanted attention from the regulatory authorities. It will sound alarms when it first appears, and will be logged on the national lightning detection network. You will have nowhere to hide.

No no. I meant that the design in itself is very clever, but it's not applicable in my case. I will have 1 spark switch (of some type yet to be decided), 1 capacitor, and 1 inductor, no antenna. Only a brief pulse of B field in the vicinity of the inductor, very little E radiated, also thanks to the very low voltage in comparation to the Marx design (in my case the E field is an unwanted energy loss that I try to limit). But I would love to know whether there is one of those out there, in working conditions. It's an old design I see, from 1960... perhaps some museum...