Re: RFI source tests...
Excellent work and insights! The root of the problem is most likely the
pronounced negative resistance characteristic that occurs once the
spark-breakdown avalanche occurs - this can be of the order of a couple
of nanoseconds for a small gap. The transition has a negative resistance
V-I characterisitic similar to that of a tunnel diode oscillator, and
like a tunnel diode, GHz level oscillations are apparently the result.
It may be possible to use ferrites between the individual spark-gap
elements to slow down the dI/dt once the heavy main conduction begins,
but it's not obvious how the dI/dt can be appreciably reduced for the
discharge current associated with an individual gap's self-capacitance.
Using smaller electrodes, minimizing gap surface area, and using wider
spacing per gap reduce gap self-C, but making the air-gap smaller would
lower the maximum breakdown voltage. Since the gap self-energy is a
function of the Vgap squared, using physically smaller electrodes, with
smaller individual gaplengths and more gaps may help minimize (but not
eliminate) the RFI. Hmm... I wonder if rotary gaps generates as much RFI
as static gaps?
Commercial arc welders don't generate large amounts of high frequency
RFI most likely because the arc-generation mechanism is quite different.
For other than a TIG welder, the arc is initiated by "striking an arc" -
shorting the electrode to the workpiece - to form a high-current path
that we are the trying to break. However, the initial open-circuit
voltage before the arc is created is typically less than 80 volts, so
the "sparkgap" self energy is much less. Once formed, the arc is
stabilized by making our energy source "look" much like a current source
through the use of inductive ballasting. The significant inductance in
the current path, and the long ion lifetimes in the arc once its formed,
severely limit dI/dt.
In our spark gaps, we're starting out with a much higher voltage from a
local energy source (the parasitic gap capacitance) that "looks" more
like a voltage source. Very high current can thus flow when the gap
breaks down and current skyrockets during the breakdown avalanche. The
HV sparkgap would have a much greater dI/dt and dv/dt to shock excite
any surrounding parastic circuit elements even during partial gap
firings on a series-gap.
-- Bert --
Tesla List wrote:
> Original Poster: Terry Fritz <twftesla-at-uswest-dot-net>
> Hi All,
> Today I did more experiments with the RFI (Radio Frequency
> I used my plane wave antenna probe:
> This probe can easily reach 100MHz in bandwidth and is not confused by GHZ
> noise affecting any electronics. As I suspected, the initial spike is
> composed of very high frequency RF. The highest I see is at 300MHz but the
> probe (100MHz) and the 60MHz scope are probably losing anything higher.
> These frequencies were causing saturation of my fiber-optic probe and
> causing just a single big positive pulse (the tell-tail sign of this).
> I have now way of calibrating the field probe in this case but I can easily
> see relative power levels.
> Tank current = 14mV (current probe reading)
> Zero Crossing spikes = 30mV
> 1st spike = 750mV
> Since the power is probably proportional to these numbers squared and
> assuming the primary current is around 100 amps, the "implied" spike
> current is around 3000000 amps!! I am not sure I believe that but...
> As I saw the other night, it is not the tank circuit firing that produces
> these nasty bursts. I would see the burst if the gap would just slightly
> click but would never go into full conduction. I removed the whole tank
> circuit and just ran the gap across the neon. Guess what? The spikes are
> still there in all their glory!!! Removing the wiring cuts down on the
> transmitted signal a little but the vast majority of the spike is still
> there just as before!!
> So it is not the main conduction that causes this RFI. It is just a little
> initial burst of little power that does it! It is just a little, hard to
> see, arc click without the primary circuit, but the RF created in that
> instant is enormous!
> So.... time for this minute's theory...
> I think the gap electrodes form a capacitor that stores capacitive charge
> energy between the electrodes. When the gap fires, this few hundred micro
> Joule of energy is shorted with extremely low resistance and low
> inductance. The air molecules must simply spray all their electrons around
> like mad in this instant. The high frequency and high power RF is created
> as the free electrons race back into orbits or whatever this plasma does.
> Once the air is ionized, the gap acts like a conductor and all the stuff is
> far far less of an issue... Making are big metal low resistance gaps is no
> doubt perfect for generating these spikes.
> So it is no wonder my previous efforts at RF filters and such we useless...
> In a conventional gap, the capacitance between electrodes is pretty much a
> given. I have no idea how you could eliminate it. If you could add
> resistance to the arc, the oscillation would be damped greatly. Perhaps
> some gas or different atmosphere... ANY inductance would drop the
> frequency like a rock and reduce the RFI level. Perhaps ferrites beads or
> something right at the gap tips...
> What one really needs to do is SLOOoooW down the initial gap firing. The
> gap needs to start conducting gently rather than in this terrible plasma
> explosion. This system would also help zero current spikes for those that
> have trouble with that too. However, a small inductance in series with the
> tank circuit my help that...
> So there are three RFI sources form just the tanks circuit.
> 1. The initial burst as the gap electrode's capacitance is shorted with
> very low impedance.
> 2. The zero crossing spikes where the already ionized gases seem to keep
> the capacitances from recharging very much on thus making them much smaller
> (but the same basic process).
> 3. the normal few hundred kHz from the yank circuit's normal operation.
> The last source is the only one that does us any good. The others serve no
> purpose and just waste power in the gap. Even though the zero crossing
> spikes contribute far less RFI than the initial spike, I think they eat up
> far more power. Thus, causing the linear decrement and all that...
> Any comments, ideas, and suggestions are very welcome. This high power
> microwave stuff is way out of my area. I would think one could put the gap
> and all in a big faraday cage (with small mess), but Richie tried that and
> it didn't seem to help. The wires leading from such a cage may conduct the
> RFI too but "good" filters may stop that.
> If these initial and zero crossing spikes could be reduced, perhaps it
> would greatly help RFI problems. Apparently, output arcs have enough
> resistance and capacitance that they are not a big RFI problem. A Tesla
> coil may be made very quiet indeed if this could be solved.
> BTW - I was wondering why commercial arc welders would not have the very
> same problem. I assume the low voltage they use stores far far less energy
> in the capacitance between the electrode and the work (like 1/500000).
> This when the arc is established, there isn't the "great" capacitive energy