TC as pulse transformer

From:  Malcolm Watts [SMTP:MALCOLM-at-directorate.wnp.ac.nz]
Sent:  Sunday, June 07, 1998 4:35 PM
To:  Tesla List
Subject:  Re: TC as pulse transformer

Hi Jim, all,
               A few comments:

> From:  Jim Lux [SMTP:jimlux-at-earthlink-dot-net]
> Sent:  Friday, June 05, 1998 2:34 PM
> To:  Tesla List
> Subject:  Re: TC as pulse transformer
> > > At some point, particularly as the Ctop gets big, can't the TC be
> > > considered as a air core pulse transformer.
> > >
> > 
> > A closer model seems to be a dual-resonant air-core transformer because
> > of the way energy actually transfers from the primary to the secondary
> > over a number of half-cycles of Fo. 
> Only if k is low. Why not make k=1?

Because it requires that clearances between the primary and secondary  
are zero. The coils would have to be coaxial and wrapped tightly 
together. Commercial air core pulse transformers go to k = 0.6 by 
running the primary as a conical winding almost the length of the 
secondary. Besides, k=1 is an unachievable ideal, even when a core is 

> > > You charge up Cpri, then discharge it into the primary of a transformer
> > > (which happens to have inductance Lpri). The flux of the primary is
> > > linked to that of the secondary, so the current in Lpri induces a
> > > current in Lsec. That current charges Csec (=Cself + Ctop), which then
> > > breaks down the air dissipating the energy.
> > 
> > Yes, but primary-to-secondary energy transfer is not nearly complete in
> > one quarter of a cycle due to the relatively low coupling coefficient in
> > a 2-coil system. Even if we stopped the proces at 1/2 cycle (at the
> > first point of zero primary current), only a relatively small portion of
> > initially available primary energy will have been transferred to the
> > secondary.
> But, you don't need to have a low k. That is just an artifact of how
> TC's have been built in the past, probably due to quenching
> requirements, etc.

Not AFAIK. Insulation between the windings is the big factor. 

> > >
> > >  Assume for a moment that your spark gap can act as an ideal switch. You
> > > close the switch, and at the perfect moment (i.e. when the capacitor
> > > voltage is zero, and the inductor current is maximized) you open the
> > > switch. All that energy has to go somewhere, and the somewhere is the
> > > secondary. Isec = Ipri * sqrt(Lpri/Lsec). (It is, of course, quite a
> > > trick to open a sparkgap switch when the current is at a maximum).
> > 
> > It's not clear that this would be the perfect moment, since only a small
> > portion of the primary circuit's energy will have been transferred to
> > the secondary at this instant. If we were somehow able to open the
> > switch, we'd develop a huge primary voltage spike from the Ldi/dt from
> > the rapidly collapsing primary field.
> But the di/dt is exactly what you want, that is what produces the high
> voltage on the secondary. You want that energy to couple into the
> secondary.

Unfortunately, with loose coupling, most of the primary flux during 
the initial 1/4 cycle is still coupled to the primary. Experiments I 
did with a MOSFET gap make this abundantly clear.

 >  Unfortunately only a relatively
> > small (10-25%) portion of these flux lines engage the secondary - the
> > remaining flux lines collapse _only_ back onto the primary, resulting in
> > most of the original energy being dissipated as heat, light, EMC, and
> > "stranded" capacitor charge in the primary circuit. 
> If only 10-25% of the lines engage the secondary, then the k is low (.10
> to .25??). Why not build a coil with a k of 1, which is an impulse
> transformer.
> >Practically speaking, [barring a single-shot explosive-type opening switch] it's not
> > possible to open the primary circuit at a current maximum with any
> > switching devices typically available to coilers, including
> > semiconductor or gas tube devices.
> Well, of course, that IS the problem....
> > 
> > 
> > >
> > > Returning to making big sparks. We know from laboratory research that
> > > making a big spark requires a slow rise time voltage pulse (many, many
> > > microseconds, if not milliseconds). We also know that we want to get
> > > energy from our storage reservoir (the primary cap) into the
> > > transformer, and then open the switch. This is starting to look like the
> > > desired switch is something that is unidirectional, reasonably fast and
> > > low loss. Perhaps a thyratron or an SCR?
> > >
> > > We want a slow rise time (necessary for developing a big spark), which
> > > means low resonant frequency in the secondary, and high inductances for
> > > both the primary and secondary.
> > >
> > > This is what Terry and Dave have arrived at, although by another route.
> > > They advocate high inductances, small C, and high primary voltages (to
> > > get the energy up).
> > 
> > I disagree with the part about a unidirectional switch. With a loosely
> > couple system, the main switch needs to be _bidirectional_ so that
> > energy can transfer to completion over several half-cycles, and the
> > switch must then be capable of being turned off once all the energy has
> > been transferred to the secondary system. This will "force" a perfect
> > quench, independent of variations in spark-loading. Heavy toploading, a
> > lower Fo, and relatively high (>=400 BPS) rep-rates will make big
> > sparks. 
> However, this would seem to be more in the fashion of generating
> > repetitively arrested streamers, each one following part of a weakenned
> > trail blazed by its predessor, with far ends "blindly" searching for
> > ground.
> It is unlikely that a streamer can repeat the path of a previous one.
> The ion recombination time in air is very fast (on the order of
> microseconds). Also, the spark leaders in a typical tesla coil are small
> enough in diameter that they cool almost instantaneously after the
> current is removed.  More likely, the apparent behavior is due to the
> sparks tending to follow Efield irregularities, and in a statistical
> sense, they tend to always go in the same place.

I have noted in a past experiment with one system that increasing BPS 
from 1 to about 3 nearly doubled the spark length. That has to be 
mostly a hot air effect. Rarefaction might be an issue. Exotic 
molecules might be another.  These were attached single channel sparks 
to a ground rod.

> In the case of lightning, you do see current flowing through the channel
> in between strokes,and the strokes making up a flash are essentially in
> the same place. However, lightning dissipates something in the area of
> 100 kJ/meter, which makes a fairly large column of heated air, which
> tends to cool more slowly than the very thin sparks of a tesla coil.  It
> is the classic surface area/volume thing.
> In some exploding wire experiments, I (and others) have created multiple
> sparks in the same ionized channel, but my energies were in the 10's of
> kJ/meter range, and my "strokes" (to use lightning terminology) were
> separated by a millisecond or so.