User-agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:24.0) Gecko/20100101 Thunderbird/24.2.0
Em 31/01/2014 07:44, Udo Lenz escreveu:
Running the coil as you do, i.e. the driving frequency close to the
secondary
resonance, probably is the most effective way of transferring energy
from the
bridge through the primary into the secondary.
I can see this operating the system with an incandescent lamp in series
with the power line (that I use in limited-power
first tests to avoid disasters). Varying the driver frequency I get
longer streamers and less light in the lamp with the
same adjustment.
Capacitive arc loading will lower the secondary fres, though, and move
it away from the
driving frequency. That causes a more inefficient power transfer to
the secondary.
Inefficent in the sense, that more primary current is needed to
transfer the same
amount of power. The secondary will then draw less power from the primary
so that more power is fed into the primary tank than going out of it.
This will cause
the primary current to rise.
Agree.
Most DRSSTCs aren't run this way. Usually the primary tank is tuned
quite a bit lower
than the secondary and since they often use zero current switching on
the primary,
the running frequency will be about primary fres.
Between the resonances there is a zero-current switching point too, that
would be the frequency
for zcs even with heavy load, were not for the capacitive loading. So
far I am insisting on no
feedback, operating the system with waveforms similar to a regular Tesla
coil, starting with a rise without much
loading that the tuning used makes as fast as possible, and then keeping
the burst at the
same frequency. This would be the ideal, were not for the extra
capacitive loading of streamers.
I want to see by how much they detune the system.
Since the runnig frequency is much less than the secondary fres,
primary current will rise for
some time almost without beats until the arc breaks out. This will
then lower secondary fres,
which moves its frequency closer to the running frequency. The power
transfer to the
secondary will be enhanced, which will grow the arc and in turn lower
secondary fres more.
Basically this is positive feedback loop leading to a rapid discharge
of primary energy into
the secondary and then to the arc. During this surge the power
delivered to the arc can be
much larger than the power the bridge supplies.
The concept seems reasonable.
I believe it is generally preferable to use longer bursts. The power
delivered by a bridge
is proportional to the primary current I, while the losses in the
transistors are proportional
to I^2. So a longer burst with less current will stress the fets less
than a shorter one of the
same total energy. On the other hand, an arc of longer duration will
consume more energy
without necessarily getting longer.
The scheme above, though, allows to make the arc time, i.e. during the
postive feedback
surge, to be made much shorter than the the burst time. This resembles
the operation of
an SGTC: The primary is slowly charged up but instead of a gap
triggering the arc,
the breakout of the arc triggers the discharge of the primary tank.
For long bursts feedback is probably necessary. In my no-feedback coil I
see that the current rises
above the designed limit of the first current beat when streamers
appear. A combination of resistive
and capacitive loading (detuning), probably.
My idea of a "perfect DRSSTC" would be one using a large primary
inductance, which
could store a lot of energy for a given max primary current and a
corresponding long burst
time to charge it up. That would reduce stress on the transistors.
Most of this is
unexplored territory.
Larger primary inductance reduces the primary current, but reduces the
maximum output voltage
too. Probably, once a certain voltage is achieved bursts length can
control the streamer length.
Again the question that I never saw properly addressed of how much
voltage you need at the output