[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Terry's DRSSTC - Controller Experiment

Original poster: "Antonio Carlos M. de Queiroz" <acmdq@xxxxxxxxxx>

Tesla list wrote:

Original poster: Terry Fritz <teslalist@xxxxxxxxxxxxxxxxxxxxxxx>

Yes, but I don't think this makes it any more efficient at spark production.
The reason being that streamer loading throws all the theory out the window.
Streamer loading is really not super heavy compared to the coil's own impedance. Losses will cause variation from Antonio's models some, but I think those models will still be fairly good. I think given a choice, it is best to try and make a magnifier that follows Antonio's models for best transfer. I certainly don't see how it could hurt. Antonio's models are just about all we have to go off of anyway ;-)

With losses, there is a continuous transition from the pure lossless system, that only charges the output capacitance, to an impedance matching network that delivers continuous energy to the output load (streamers). The waveforms that Terry shows below are good examples. Without losses, with my tunings, the energy transferred to the output return to the power supply if the driver continues to operate after the maximum output voltage is reached. With small losses, just a fraction returns, and after a certain loading energy is transferred to the load continuously.

Antonio's theory assumes that breakout doesn't happen until the very end of
the burst. But with the current trend for big sparks out of small coils,
breakout happens early. Hence the total energy delivered to the streamers
can be maybe 6 times more than what the topload can hold when charged to its
breakout voltage. This is a radical departure from traditional TC theory and
you need to think about it for a while. But I believe it is true for DRSSTCs
and probably for high powered spark gap coils too.

The coil is only really lossless for maybe 80% voltage on the very first bang. After that, the streamers will reduce the arc out voltage drastically. The DRSSTC can certainly drive power for as long as it want to during a bang since it is fed from an infinite source or until the IGBTs blow up. It might be a "next step" to try and "modulate" a DRSSTC to give a constant voltage output for say milliseconds. If you stay within the IGBTs "normal" load range, the voltage could be maintained indefinitely. Maybe like a single bang that lasts two days ;-))

I really doubt that significantly more energy than what can be stored at the output capacitance can be fed to the streamer load, before the system blows out due to excessive input current. But this is, in principle, possible, as the system can be designed as an impedance matching network. Interestingly, the design resulting from this is practically identical to one of the lossless designs. Again, the driving frequency must be between the resonances, otherwise the input current and the energy stored in the primary circuit grow explosively.

If I understand right, a coil designed to Antonio's theory can never produce
a bang energy bigger than the charge the topload can hold, so it would look
pretty wimpy in comparison to the tabletop coils Steve made that spark 3
times their own height :-o That is why we tend to stay with self-resonant
operation at the poles.

Antoinio's stuff never considered a none energy limited power source (who would have ever thought...). However, that should not be too hard to reconsider now that the DRSSTC is here.

My drsstc design considers an ideal voltage source as input, that can deliver unlimited energy.

Anyway, once the secondary is loaded by streamers the whole behaviour
changes. With a heavy enough load the two resonant peaks get damped out and
the whole Antonio vs. Ward thing becomes a non-issue- to keep the sparks
growing you just need to drive a frequency somewhere within the "bandpass"
of the "filter".

I agree. But both designs, for lossless operation, and as an impedance matching network, happen to be practically identical.

So, you could have a "hybrid" mode that uses Antonio mode to ignite the
initial breakout and then continues driving to make up any desired bang
energy. The resulting waveform would have a single notch followed by a flat

But the control circuitry to do this reliably with feedback is difficult
(you may need to change the feedback phase mid-burst) I have done it on the
bench without feedback but the tuning was incredibly sensitive and it only
took the slightest change to make the notch disappear. So I guessed it would
be too sensitive to streamer loading and went back to the plain
self-resonant approach.

This is what my DRSSTC does all the time:
In the top waveform the current normally switches every 5uS. But in the middle there is a 7uS pause. In the lower waveform I have a ~15uS pause!! Goodness knows what the phase and frequency/time shifting looks like. But I use pure current to control the timing so it is not a problem in my case.

Good demonstration that the lossless design works well with loading too.

Antonio Carlos M. de Queiroz