Spark gaps -- making and triggering

Win, thanks, I did not realize that there were HV relays and did not think aboout high power HV resistors. I will check it out... Thanks a lot...

i

According to Ignoramus29428 :

Hmm ... I remember a high voltage test cage which I used to have to calibrate at a place which made high-voltage rectifiers (among other semiconductors) was set up in a metal electronics rack with a door of

1/2" Plexiglas. The door was held closed by a long screw cut with a buttress thread (lots of slope on the side towards the free end, and almost vertical on the side towards the knob which turned it. The female thread was a hole in two plates with a spring wire between them pinching on either side of the screw.

To close it -- all you had to do was push.

To *open* it, you had to spend a couple of minutes turning the knob to unscrew the screw -- and about halfway during the time a cam would swing a safety grounding bar into contact with the high voltage terminals, to make *sure* that there was no high voltage there when the operator reached in to change the device under test.

The larger the radius, the shorter the gap will need to be, because charge concentrates on small radius points. I've got no idea whether the TIG electrodes could handle that much power (but I'm not sure how many caps you intend to hook in parallel for this.)

Does argon have a lower dielectric constant or breakdown voltage than air? I would expect it to be somewhat more insulation than air.

I think that you'll have to inject a very high voltage pulse between the two with the electrode closer to the ground side, but capable of ionizing the air to make the gap enough shorter to dump everything from the caps -- with something like a flash lamp trigger transformer. (Hmm -- perhaps a spark coil from a car?) Part of the problem is keeping the trigger transformer from being a consumable. :-)

Perhaps set the gap small enough so at near your full voltage it will break over on its own, and simply set it up and then start charging, letting it trigger itself -- though you won't know exactly when.

Perhaps UV illumination will ionize the air enough to trigger the gap.

A good idea -- this strikes me as a very dangerous experiment. (But possibly fun. :-)

There is no way I would generate that much of an EMP near all of my computers. :-)

Enjoy, DoN.

Ignoramus29428 wrote: ...

Thoriated tungsten is used in some spark gaps - the slight amount of radiation making the trigger level more consistent. (Beyond that interesting item, I know nothing else about tungsten in spark gaps)

Probably safe but not effective. However, here's an argon-based method I thought of that might work, with some work: Put round metal contact plates at opposite ends of a sealed cylinder (say 1" across and 1" apart). Through one valve, fill with high-pressure argon before charging caps (or at least before arming the discharge circuit). When ready to fire, exhaust the argon via second valve into a several- gallons evacuated tank to rapidly drop the pressure. See Paschen curves (discharge V vs. P*Distance or vs. P) for argon and for air at

eg.

Also see

a low-pressure spark-gap switch triggered by light or by an electrical trigger. It says, "The neat part is that a spark does not form. Instead, a diffuse discharge occurs" and "A simple switch based on [figure] can operate at potentials of 35 KV and handle currents of 10 KA with rates of current rise of 4 X 10^11 amps/second.", then refers to a Maxwell Labs patent for 1 MA "Coaxial Pseudospark Discharge Switch" and Tetra Corp patent for 50 KA "Linear Aperture Pseudospark Switch".

-jiw

["Followup-To:" header set to sci.electronics.design.] On Wed, 05 Apr 2006 01:16:10 GMT, Ignoramus29428 wrote in Msg.

If I wanted to build this I'd look into using a small pneumatic cylinder to bring the electrodes together.

BTW, a setup like this is really more like a relay, not a spark gap. It's a relay that happens to spark a lot ;-)

robert

["Followup-To:" header set to sci.electronics.design.] On Wed, 05 Apr 2006 01:55:09 GMT, RoyJ wrote in Msg.

Why not bring them together all the way? To avoid them welding together?

robert

can crusing coils will try to flatten themselves. They don't explode.

Trying to contain giant bangs in plastic boxes is called making a bomb. Although a bit different, it's the same reason gunpowder is shipped in really flimbsy containers, not tough containers.

That should work fine. Or you could use copper or aluminum flashing cut into wide strips, or extruded aluminum. I wouldn't recommend using steel in the high current path because of high skin effect losses.

3/4" is pretty small - something like 1" - 1.5" might work better for a variety of reasons.

Using 1" pipe end caps should work fine.

By using thick wire or tubing, a large diameter coil, and applying limited energy, there's no risk of the work coil exploding. There are radial "magnetic pressure" and Lenz's Law forces that try to expand the coil's diameter, and there are axial forces that squeeze the turns of the coil together. All of the forces act in a direction that tries to increase the inductance of the coil (i.e., increase diameter and decrease axial length).

If you use heavy gauge copper wire (#6 - #8 AWG) or 1/4" copper tubing and 3 - 5 turns, the coil will have sufficient mechanical strength. If you use copper tubing or bare copper ground wire, you'll need to slip some heat shrink or vinyl tubing over it before winding. Using a space wound coil using bare wire will likely fail since axial forces may close up the inter-turn gaps, shorting out the coil.

It doesn't seem to have any effect, the gap simply fires as the electrodes are being brought together. I do use a tightly twisted pair of wires going to the solenoid - this helps to prevent any induced EMF's from the pulsed magnetic fields in the area from creating any problems.

The solenoid is electrically isolated by distance and by using good insulating materials. You could use an air cylinder to actuate it instead.

Sorry - the correct URL is: http://205.243.100.155/frames/Newgap2a.jpg

Thanks!

Bert

These should work fine - relatively inexpensive and easily replaceable. I use silicon bronze bolts for bolting the work coil to the bus bars, and the threads have lasted for thousands of shots. Just make sure that the threaded section is locked down to prevent arcing.

Bert

------------------------------------------------------ We specialize in UNIQUE items! Coins shrunk by huge magnetic fields, our "Captured Lightning" Lichtenberg Figure sculptures, and Out-of-Print technical Books. Visit Stoneridge Engineering:

Win,

Are you using high energy bulk resistors (such as Kanthal, or HVR Advanced Power) to rapidly absorb the energy? I've seen high power wirewounds fail spectacularly under similar conditions.

BTW, HVR has a nice thermal modeling tool for designing rapid energy dump solutions:

Bert

Although the BANG from a high energy air spark gap is extremely loud, it doesn't generate volumes of high pressure gas like chemical explosives. The container must be capable of handling the shockwave and acoustic pressure. A number of manufacturers (General Atomics, R. E. Beverly) currently use Lexan polycarbonate housings for their high energy spark gaps.

Bert

Great!

Good, then, that's what I will do.

That sounds simple enough. I will indeed use copper tubing, I think.

I appreciate your advice.

As for triggering the gap, at first I could try simply setting a good distance. That would not get me all the way up to 13 kV that my power supply could make, but maybe I can get close.

i

A properly adjusted fixed gap will work. However, you may see significant shot-to-shot gap breakdown voltage variation. Personally, I prefer to be in control as far as WHEN the spark gap will fire. I really don't like sudden (loud!) surprises in the high voltage lab... :^) http://205.243.100.155/frames/kaboom.jpg More importantly, you should plan to disconnect your charging power supply before firing the device. If the gap happens to fire while charging you run the risk of blowing up the HV diodes in the charging supply. Also, consider connecting 10-20k current limiting resistor(s) between the charging supply and the capacitor bank. These should be

100-225W wirewounds for sufficient worst case standoff voltage. This will limit forward diode current during ringing, and should prevent backfeeding high current back into your charging supply if you happen to encounter any HV diode failure(s).

If you ground the charging input after firing, these resistors can serve as bleeder resistors to dissipate any residual bank energy, or for an emergency shutdown. Similar to Win, I leave a set of bleeder resistors across the bank at all times (other than charging) to prevent any surprises.

Did I mention, I really don't like surprises in the HV lab? :^)

Bert

Thanks... I went to farm supply and bought some 1/4" copper tubing, 1" by 1/8" aluminum flat to make a bus connecting several caps, 3/4" copper pipe and corresponding pipe caps. I will start putting a plan together.

i

You do need HV relays, which can be expensive. I get many of mine from eBay sellers. Manufacturers are Kilovac, Gigavac, and Jennings, which you can use in your search terms. You can expect to pay $75 to $250 for a 30kV vacuum relay, and perhaps more if I'm bidding against you. :-) I just won three Kilovac KC-38 25kV spst relays for $60 each, a better price than usual.

Here's a nice 35kV spdt relay with cables and HV connectors.

At modest power levels I use series stacks of many 2W carbon- composition resistors. These dissipate the energy throughout their bulk, rather than just in a thin wire, etc. For higher power levels I'm building up a collection of Advanced Power resistors, mostly taken from high-power IGBT snubbers, etc., but quite a few of these will be required in a series stack.

Thanks, Bert, that's a good lead, for repetitive applications. I've created simple heat-spreading spice models of transient- energy absorption in the 0.1us to 10ms timeframe, which I've vetted against microsecond-scale measurements. You know, cells of thermal-mass with thermal-conductivity connections. With this I was able to see how well-designed TVS parts are - I was impressed. I've also studied MOSFET avalanche with these spice models. We have a site-license for a good FEA program at the Institute, but I haven't taken the time to learn how to use it.

Thanks Win, I will keep an eye on eBay for HV relays, in fact I set up a snipe for one.

i

My 100uF 10kV and your 70uF 12kV caps each hold 5kJ at full charge. The largest HVR C-series resistor, C2654, about 1" in dia and 2" long, takes 5.6kJ for a 125C rise, and is rated for 22 to 44kV operation, depending on the resistance value.

Here's one on eBay, Too bad he only has one resistor. But his is a pretty high resistance value anyway, 2.8k, which has a 280ms time constant with 100uF. HVR says you can use 5 ohms and still apply up to 22kV pulses. That would be a 500us time constant with 100uF. But 5-ohms would also create a 2000A peak current with 10kV, compared to a 400A typical max spec for Kilovac relays. 100 ohms would mean a 100A peak current and a 10ms time constant.

I'd rather use two C2654 resistors per HV capacitor. Two 1k in series would give a modest 5A peak/cap for long relay life. That'd be 8 resistors and 20A peak for a bank of four big caps and take one second for five time constants (10kV to 67 volts).

HVR's rod and tube style is attractive, the 10" ones handle 22kJ, so only one or two would be required for a full 4-cap bank. Actually, I have a modest selection of these in stock, from the Flea at MIT.

You can easily build your own as well. For example, here's a design from HV hobbyist Bob LePointe:

You can also get some potential construction ideas by looking at some of the open frame HV relays made by Ross Engineering: Bert

Bert, thanks, Bob's relay looks very nice. If I do not win the relay that I set a snipe on (I usually lose on most of my auctions), I will make something like what he did. I believe that I have a 12v pull solenoid lying around somewhere.

I think that it makes sense for me to make a system where it is electrically impossible to have a power supply connected to the system, at the same time as the spark is triggered.

i