High energy pulses and power switching

[ProdigyChild]

This thread is meant to cover theory and practical plans of devices for creating high energy pulses with extremely short power rise times. Preferably, triggering of the pulses should be possible by a low power signal (electric or other).

Such technique can be used to set off high explosives and do lots of other unuseful yet fun things :D
Explosive chemicals as the most obvious approach are ruled out.

Charging, storing and switching of energy/power are the challenges to take.
Easily obtainable materials should be used.
Understanding and imitation is more important than someone boasting of the very special device he just got from e bay (and no one else will be able to do).

Probably, most of the time we'll discuss about high voltage generators and spark gaps - stuff you can abuse for blowing EBW (exploding bridge wire) and EFI (exploding foil initializers) kind of toys.



Guidelines for this thread:

a) use the title field to summarize, what you gonna talk about.
b) whenever possible, try to add quantitative data rather than suspicion. '1.8kV' is much more expressive than 'high voltage'. If unsure about a number, let us know, you're unsure. It's not a shame to be unsure, it's a sign of awareness!



As a appetizer, here's a link to get an idea about industries state of art:
http://optoelectronics.perkinelmer.com/content/RelatedLinks/HESbrochure.pdf
We can do most of these things, too :D

Note, there's already a thread about EBWs, if you're simply interested in getting a EBW-cap to work:
http://www.roguesci.org/theforum/showthread.php?t=3425

...and the water cooler 'bridge wire project':
http://www.roguesci.org/theforum/showthread.php?t=4339

[ProdigyChild]

Glass has a very interesting property: it is purely elastic! Either it goes back to its original shape after removing shape-changing forces or it has already broken. At least if the time scale is not centuries.

Glass can have enormous tensile strength: S=3400MPa (=3400N/mm2) for E-glass fibres and can be stretched by e=3.5-4% until break. Elasticity: E=73GPa (similar to Aluminium). That's at most W/V = 1/2*E*e^2 = 58.4J/cm3 energy until break down.

That allows to construct an interesting device:
a glass rod + a piezo loudspeaker connected to the middled of it. Lift a ton of material with that rod. Then connect a function generator to the piezo and excite the glass rod a resonance frequency......clink......BAM + ground trembling!

Switching a few kN with a battery powered electronic device and NO explosives and NO human-detectable sound (ultrasound) :D

Possibly the same technique is applicable to ceramics.

Pure fiction?

[Marvin]

In the glass rod method the piezo idea sounds over complicated and a little unreliable anyway. How about simply heating it rapidly with a coil of wire causing it to crack, or if that were too unreliable, heating a metal rod in contact with it forcing it against the rod. Whatever method I come up with, I still cant think of anything that would be an improvement over 2000 year old egyption or greek technology. For example a trap door closed with a lever and secured with a pin. Tiny amount of force needed by the pin to keep the lever in place, it goes the trap door opens, or 2 chain ends held with a link that swings open or many other reliable methods.

Spark gaps, Marx style switching and thyratrons are almost certainly the best way to go for easy high power pulses, but they tend to fry semiconductor circuits. The use is very simple I'm not sure what you are hoping to learn. Setup a spark gap thats just too wide to fire and find a way to make it fire, introduce ionisation (an interelectrode, UV flash, that sort of thing), or a sharp voltage spike (say from an electrically connected second spark gap at lower voltage), or alter the distance (say by having a spinning inner electrode) and it conducts.

With regard to fast storage if low tech rules and explosives arnt an option stacked foil capacitor are pretty much the pinnicle and the only question is the plastic. Teflon rules, PE will do, PVC pretty crappy, PP good. (all from memory so not 100% reliable.)

Most of these electrical topics have been done to death on the internet allready, mainly in the area of tesla coils, theyre cheap to do, they work but operation depends a lot on the construction. If you expand these basic things into semiconductors you have the oppasit features, operation depends less on the construction and much more on the design, its much more expensive and understanding the circuit properly involves 90% of the subject of electronics.

[ProdigyChild]

Child-of-Bodom uses a car relay to switch a capacitor. The contacts of the relay weld together while switching.

This morning I had the idea of modifying the relay as follows:
Stop the contacts from touching. Let them only approach each other to a distance of save SBV (self breakdown voltage) of an air spark gap. When the relay 'switches' an arc is formed that bridges the two contacts. Welding contacts is impossible.

The distance for a 2kV arcover should be less than 0.4mm according to my experiments. So a bit precision will be required for that modification.

[Child-of-Bodom]

The welding is no problemo, I just make a small hole in the casing of the relay, and with a screwdriver you can open it agian. The relays won't last forever this way, but how many times do you actually use them :rolleyes: ?

I asked Perkin Elmer what a Triggered Spark Gap would cost, the smallest one, 2-4KV, 10KA -720 Euro per piece :eek: :eek: :eek: :mad:
So that is a no-go.
I like the idea of a TSG, but what eats me is the fact that a simple drop of water, or a higher humidity can make the whole system far from safe. What would be the minimal amount of times the gap should be larger then the auto-breakdown distance? 2 times looks a little less, 3-4 times? Does it still work then?...

[Anthony]

What about a solenoid actuated switch?

This one doesn't look too hard to make, and is supposed to be good for 50 000J

http://205.243.100.155/frames/gallery/newgap5a.jpg

http://205.243.100.155/frames/Newgap2a.jpg

Source: http://205.243.100.155/frames/shrinkergallery.html Bert's quarter shrinking page.

[ProdigyChild]

COB: If you avoid the welding, it will last forever!
I estimate I have 30+ switches per evening :D

Provided, the electrodes are big enough, you can hardly detect any wear of the spark gaps. I made lots of experiments with a vacuum spark gap (0.1atm) and fired it a few hundred times in short circuit mode. The copper electrode plates (5mm x 5mm for the arc, 10mm x 10mm for cooling) show no wear/oxidation, they're only slighly lighter in color. Glossy surface!
The arc of this type is about 8mm in diameter and 20mm long. I believe, the low I/area avoids wear of the electrodes. I like the faint sound and bright flashing of it.

I haven't build a ambient pressure spark gap with suitable diameter so far, only max 2mm diameter rods. These do have wear, but still good for umpteen shorts without much change of SBV.

Perkin Elmer do make profit with such prices!

Humidity, water?? That's why we should build a case for our equipment....

About breakdown voltage: The document above says: operating voltage = 25% up to 80% of SBV. You can't go below the 25% because triggering becomes difficult and unreliable. Double the self triggering distance will be enough!

Above all, a high voltage trigger pulse is required. I've tried to trigger a 2 kV gap with voltages up to 1500V (BU205 transistor maximum voltage). It was a failure! Sometimes works, sometimes not, sometimes self triggering. Even high current pulses with small caps are by far outperformed by a low current high voltage trigger. I got the impression: once you have a suitable trigger voltage, the design of the spark gap does not matter too much. Almost everything involving three electrodes works.

I'm designing a PCB for my (future) current-pulse-box. It contains a stabilized high voltage charger (400..2000V) + 400V - SCR-switched trigger generator using a clumsy and heavy car ignition coil. I hope I can put the spark gap on board, too, without killing the semiconductors from the current peaks. Probably finished this week-end! I can't wait to determine trigger-signal-to-fire delay!
I expect ....errr.... hope ;) less than 20us.

[Marvin]

Decouple the pulse electronics from the semiconductors as far as possible. Spark gaps cause high voltage ripples in the ns range on ground lines and these can fry semiconductors.

It sounds like you are generating a high voltage and switching this with a semiconductor direct the trigger terminal. Thats.... Umm.. Not ideal.

Is firing delay a critical factor?

I'd be inclined to suggest something more like you find in a camera flash circuit which is essentially a shorted thyratron. What you are making is an air thyratron. A second tiny cap is charged to say 300v and this is shorted into a 20:1 coil with about 10 turns primary. This induces very quickly about 6kv over the secondary which is connected to the strip outside the flash tube inducing the ionisation and it fires. My point is that the switching mechanism for the trigger never needs to deal with the triggering voltage of 6kv, only the primary voltage of 300v. A simple pair of contacts are all thats needed there, its fast and reliable and 300v isnt much for a decent semiconductor to handle if absolutly needed. You will need to check impedences though. Its just a shame the power in the coil of a camera trigger xf wouldnt be enough to trigger an air gap (I shouldnt think).

Why are you aiming for less than 20us and why wouldnt mechanical contacts do for the firing button?

[Zerstoren Sie]

Anyone ever try a mercury switch?? Basically one simply dips one contact into a small pool of mercury. The switch time is extremely fast and there is no bounce. (Thus, it can also be used to manually switch logic circuits) Assuming you could find the mercury, all you would need for a high power switch is an enclosure and a way to move the contact into the mercury. Nikola Tesla's patents cover this technique.

Here's a list of them: (U.S. Patents)
609,251
609,245
611,719
609,246
609,247
609,248
609,249
613,735

He calls it an Electric-Circuit Controller. All of these use a rotating contact for fast switching. They should provide ideas though for ways of going about this. Maybe a solenoid with large contact attached to it would work.

[Jacks Complete]

Zerstoren Sie,

Mercury vapours/fumes kill, and mercury is actually quite tricky to deal with.

If you want to go down this road, make the entire switch outside, then pump the air out as best you can, flush with Argon, pump down, flush again, fill with argon, and then reduce the pressure to probably half an atmosphere, then seal it. Toggle it remotely, perhaps by tilting, since the mercury will run to the lowest point, behind a screen. Put it on a tray so that if the glass breaks you can catch the expensive mercury and not have it soak into the carpet!

Tesla never knew mercury caused madness, and he was, shall we say, not completely sane by the end...

An air switched, pump down trigger would be easier and safer to use. Just get your electrodes, put them inside a glass tube, about twice as far as you would expect it to arc in still air, and leave a small hole past one, for the air to get out. Hook it up to the vaccuum pump, then when the pressure drops far enough, you will get an arc. If you can't get the vaccuum low enough, just turn it all off, and shift the one elctrode further towards the other. Ensure the pump end is the grounded end, btw.

If you are going to use the UV output for anything, carry on. If not, sheild it with enough that you can't see the arc, since it will sunburn you and give you bad eyes if you don't!

[Boomer]

We are advancing in technology! I nearly feel ashamed that my EBW box still works by touching the second cable to the hot terminal. :o

Have you actually measured the current rise time? I once used a flash tube as a switch, triggered by photons from a second tube which again was triggered normally. This was for insulation mostly. I had to move to a monster relay (8000V/4000A, self made from copper bar, iron + big e-magnet), because the tube had a current rise time of several µs, while mechanical switches get to 500ns or below (limited by circuit inductance).

I do remember that arcing on HV lines takes 200-300ns for full ionisation, or so my professor said. I never tried a spark gap because of the slow arc in flash tubes. Was it the argon, the fact they work above 1 atm. or was it the photon-triggering? :confused:

P.S. Just realized that at the speed I move the wire towards the terminal, there is probably an arc before it fully touches. Meaning they CAN be fast, at least if <1mm short ( -> flash tube too long at 50mm ?). :)

[Marvin]

Flash tubes are designed to be fairly slow, and to have nontrivial on impedences so they arnt a good example. The cap in those circuits are also very slow. Most camera tubes are low pressure xenon, I'm unsure what you were using. Some professional tubes are high pressure capillary tubes, they would be even worse speed wise.

For specific applications clever designs switched by a spark gap can produce a pulse in the region of 10ns long.

For high voltages a switch is just a variable spark gap, the problem is they cant handle the same sort of current without damage, rise time shouldnt be an issue.

If you look at the links Anthony posted the stunning thing there is that though theyre using spark gaps which can be nanosecond fast the energy dump takes in the region of 100 us! Its totally determined by the rest of the circuit.

[ProdigyChild]

A hard lot of work is behind me - my power-pulse box's main PCB is finished!

Bad news first: I have really problems in acquiring numbers without a digital oscilloscope. I can't trigger from the digital input signal. The following surges disturb almost everything and cause the oscilloscope to trigger upteen times thereafter. A real mess on the screen. It's rather guessing than measuring....

Most probably the delay between input signal edge and spark gap firing is in the range of 30-50us.
It seems, the spark gap needs 2-6us (i.e. about 4 us jitter) from HV trigger to conduction. However, commutation is definitely much less than 500ns (unmeasurable with my equipment), peak current 2.0-2.4kA with 3x0.33uF @ 2.0KV.
What means, the device makes at least 4kA/us rise :cool:
It should also be noted, that the caps don't give more than 2.5kA, when shorted with a wire. This is nearly the same value as with the spark gap.

The longest time is consumed by the HV trigger. It's build with a low cost SCR (C106D) and perhaps can be improved. Using a car ignition coil isn't optimal either. I used it, because current rise and peak current are within the SCR's specifications, when discharging a 0.1uF/400V cap into this coil. Maybe I try a HV-MOSFET instead.
For most applications, the utmost simple SCR triggering mechanism is sufficient.
The data sheets for average power SCR's (<10A) promise switching times of 0.6-2us.

About the 'onboard' spark gap endurance - 80mm from CMOS circuit by the way:
I let the circuit run with a NE555 signal generator (1/3Hz) input for 1.5hours. 150uH 'load', 2J energy. That's 1800 shots! The contacts and trigger electrode become black after a few minutes, without degrade in performance nor a single misfire. It works reliable from 1.6kV-2.1kV (MAY fire at 1.2kV, but no guarantee). I should provide a photo, soon.
A well adjusted spark gap is really reliable!

1/2 hour with shorted trigger signal input and blowing at the gap, flashing it with a very powerful flash, switching on and off lights did not cause any spurious trigger. However, leaving digital input open causes spurious triggers all the time, especially if I touch the curcuit with a metal part. Not shorting trigger input will result in premature firing :(

@Boomer: when experimenting with low pressure and 10-20mm of electrode distance, I observed considerable performance loss - more energy converted into light! I haven't measured the timing, so I don't know if that was bad, too. Worth, repeating this experiment....

I've also experimented with a broken car relay. I tried to adjust it, to not close the contacts, but only let them approach 0.2mm. I failed. Normal relays are too elastic.
I had to adjust it to a such large gap (when 'on'), that no self triggering occured. So I ground the contacts off, an put a sturdy copper wire there instead. Now I have a relay, that arcs over at 1.9kV, when switched 'on', and estimated 4kV when 'off'.
Works fine, too.

EDIT: I've added photos. The capacitors are not shown. Surrounding of the arc lighted digitally to reveal details.

[ProdigyChild]

I replaced the spark gap by an improved one. The electrodes consist now of 1mm copper plates pressed into proper shape in a vice. I used a digital oscilloscope to measure swiching times and delay (OMG...):
input trigger to peak current = 8-9us, low jitter
rise time = 500ns for 2000A peak maximum

Delay is much better than expected!

Rise time is still difficult to measure, because it consists of a 'RF noise' of +-4000A. I'm quite sure, the oscilloscope or cables are a bit confused at that time.
And I used a transformer as load, rather than a resistor. I can't explode wires at work, do I ? ;)

I included a schematic of the circuit. Feel free to provide criticism, I'm not an expert, nor reluctant to improvement. I can provide a Postscript/PDF layout , if desired.

The short delay between trigger and firing allows to build things like 'stopping a bullet by a detonation in front of it'.
A shockwave travels less than 10km/s *10us = 100mm before the circuit fires after sensing a shockwave and jitter is only about 10mm :cool:

Another interesting thing to do with that circuit is to produce high voltage pulses at high current. I build an air coil transformer for 10kV / 6A+. Using 10 times the ratio, it would be possible to produce an X-ray pulse with a rectifier tube...

Spark gaps rock!