Electronics > RF, Microwave, Ham Radio

EMI from wire vs spark gap.


Say we have a loop of wire about 6 inches diameter and one end of the loop is cut and terminated to the high voltage side of a spark coil e.g. car ignition coil, and the other end of the loop is cut with a half inch gap so a spark can jump across it. Is there a great deal of difference between the amount of interference radiated from the wire loop magnetic and electric field, and how much is radiated form the actual spark? That is to say, if the gap was completely shielded from the outside world, the wires would radiate a certain amount. If the gap was then exposed, would the radiated interference increase only slightly or a significant amount. What I think I am trying to say is, does a spark of a certain length radiate more than a wire of the same length?

To where is the start of this loop connected?  Right at the spark plug, or back by the ignition coil?

Even by the spark plug, I don't think you'll get a secondary spark, with that length of wire.  Anywhere else along the spark plug cable, the sharp edge will be too attenuated to get any spark this way.  If I've understood your proposed circuit correctly.

If there's no primary spark gap at all, the result is nothing: the coil has too slow of a risetime to be able to drop voltage across a 1-2ns length of wire.

If it doesn't spark, the radiation will be entirely due to electric field from the wire, at frequencies where the coil resonates (~100s kHz).  If it does spark, there will be some RF emission characteristic of the loop, yes.

Your circuit isn't explicitly "wrong", though: a spark shorting one end of a transmission line to ground (the wire through space is a poorly-impedance-controlled, very leaky, transmission line) can induce a spark from the other end of the line, to the start end.  The resulting current waveform has a rectangular shape, characteristic of the transmission line length.

I had such a problem on a recent experiment,

This is a stack of 2N3904s operated in avalanche mode.  The bottom (far right) transistor snaps on first, discharging its ~100V drop; suddenly, the next transistor (to the left) sees 200V across it, and avalanches immediately, and so on, until the full stack is at transmission-line-output voltage (about 500V in this case).  The mica capacitors then discharge through the line, giving a pulse of a few nanoseconds duration.

Since the stack is powered by 1000V, I was having occasional sparking around the highest voltage end.  The red arrow indicates where I had to cut in more ground-to-pad clearance.

Actually this is the inverse problem, that the high voltage pad was arcing to ground, instead of the transistors discharging it.  There were a few crusty points towards the cable end, though, which I also had to clean up, and those have the spark-triggered-by-spark series of events.


The wire goes from the ignition coil secondary, half loop of 3 inches radius, half inch spark gap, then another identical half loop (makes a circle) back to the grounded other end of the coil secondary. I had a microcontroller based ignition box going on the bench and with the spark completely enclosed it ran just fine, but with the spark out in the open air the box would occasionally hiccup slightly. Bare pcb, not in a box like it is supposed to be. Is the exposed spark radiating more noise than the wires feeding it?

Oh, ok, so the spark discharges the end of a loop.  Yeah, that throws off all kinds of near- and far-field crap -- consider that the line is charged up to 10kV or more, then discharged in a fraction of a nanosecond (sparks in air are very fast!).  The peak current can be hundreds of amperes, so it doesn't take much proximity to induce dangerous voltages and currents (on the order of 10V to cause logic glitches, or 100s to 1000s of V to cause ESD transients, device hiccups, and possibly, real damage!).

It's not meaningful to ask "is the spark radiating", because the frequencies that are being radiated are not characteristic of the length of the gap (~1cm, so >1GHz).  The frequency carrying the most energy is probably in the 100-300MHz range, or one at much lower frequencies with secondary peaks similar to that.

Conversely, even if it were the gap radiating, how could you know?  You can't resolve ~10cm+ wavelength radiation to that fine of a resolution (just as an optical microscope can't resolve <200nm features), so again, it's not meaningful to ask. :)

It would be meaningful to ask in the optical range: the spark gives off everything from IR (it's hot!) to UV (it's really hot!), with peaks due to atomic transitions (especially blue from nitrogen).  But that's only where a modest fraction (maybe... 20%?!) of the energy goes.

With everything being close, it's more meaningful to think in terms of magnetic fields, loops, and induced image currents.  Anything making loops in the plane of the spark loop, will pick up induced voltage or current as well; anything perpendicular will not.



[0] Message Index

There was an error while thanking
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod