Re: [TCML] SSTC full bridge control system question

[Steve Ward <steve.ward@xxxxxxxxx>]

If you tune primary L*C = secondary L*C, then the primary and secondary
currents should be a sinusoid enveloped by another lower frequency sine wave
(aka, "beating").  You should be able to run the primary for one "beat"
cycle, which for you should be about 1/.15 (~7) cycles long.  This should
the best secondary peak voltage vs primary voltage/current.  When the
secondary is peaking, the primary is notching, because the energy has been
transferred from the primary to the secondary over those 7 cycles.

I did not realize you are attempting to get this much power from a
half-bridge, that will be very hard without a lot of current, so 1.6KA seems
like a reasonable current to get that much secondary energy.  There really
isnt any way around this, if you work through the physics, you are going to
be required to store a significant amount of energy in the primary before
the secondary can use it up, and this means high peak primary currents.
Streamers prevent you from building up energy over a really long time in a
high impedance primary (so I is less because L is bigger), so if you want
fast streamer energy delivery (which makes for efficient streamer growth)
then you need high peak primary currents.  Tuning should only make a small
improvement in spark length vs primary current, once you get it close of
course.

If you can go full-bridge drive, id suggest it because it sounds like you
are really after much more power than a practical half-bridge will support.

What silicon are you working with for the main switches?

Steve

On Sat, May 1, 2010 at 10:25 PM, Michael Twieg <mdt24@xxxxxxxx> wrote:

> Thanks for the quick reply Steve,
> I understand that 1-2MV is unrealistic.  I'm using that as a goal for
> unloaded output (no streamers), so obviously in reality a spark will
> break and compress the voltage to much lower levels.  But in my
> simulations I'm not even able to get to 400KV before the primary
> current hits 1.6kA, and this is without any load on the topload.
> We're working with a coupling coefficient of about 0.15, and our
> resonances fall around 24KHz and 34KHz.
> I definitely get what you're saying about simulation being dubious. At
> this point the coil is still being built and I want to be sure the
> controller is good to go when it's finished, so simulation is all I
> have to go on.  The coil will have a tapable primary (btw I mistakenly
> said our coil is a two coil system-it's actually double resonant, my
> mistake), and we'll be able to change its elevation to some extent in
> order to change our coupling.  So we'll have some control over our
> system, but I don't think any combination of parameters will allow me
> to operate at the secondary resonance.
> I'm not sure I understand what you mean by notching in the primary
> current.  Are you referring to the minima in primary impedance (as
> seen by the full bridge) that occur at the two resonance peaks?  I've
> noticed that a zero-crossing converter like yours seems to always
> resonate at the frequency with the lower primary impedance, and this
> is part of my concern.  If it were possible to operate at the other,
> higher impedance peak, I think it would be possible to get at least as
> much voltage on the topload with much lower currents from the half
> bridge.  That's what LTspice is showing me, anyways.
>
> -Mike
> ---
> Adjust the primary tuning for best energy transfer to the secondary.  This
> can be somewhat pointless to do in simulation, other than just getting a
> feel for how a 4th order resonance behaves to various inputs. The reason
> modeling it is almost pointless is that no one has a really good model of
> the streamer load (i have some models that are closer than others ive seen)
> which really determines largely the behavior of things.  Also, 1MV is a
> HUGE
> tesla coil output and likely unrealistic.  My big coils are about 700kV
> peak
> i estimate from base current and other simulations, and i need about
> 1600Apk
> primary current to get there.
>
> Anyway, im not sure if your analysis is right.  The double peak comes from
> the mutual inductance between the coils, where the M either adds or
> subtracts from the apparent resonant inductance, which means there are 2
> peaks, one just below Fres and one above.  Increasing the coupling (hence
> more M) causes these peaks to move further apart as the adding/canceling of
> the M term is more dramatic.  Generally, the most *efficient* tuning method
> is to tune the primary to the secondary which results in primary current
> notches.  The primary current notch is indicative of complete energy
> transfer to the secondary (secondary I and V should be peak during primary
> I/V minimum).  The transfer time is essentially controlled by the coupling
> coefficient, where it should take 1/k cycles for the transfer to take place
> (so k=0.1 needs 10 cycles, k = 0.125 needs 8 cycles etc..).  This places
> some upper bound on the energy you can deliver to the spark within one
> energy transfer cycle, for small coils it often ends up you cant get enough
> energy transferred within 8 cycles or so.  So the other trick is to tune
> the
> primary to excite just one of the resonant "poles" which means the primary
> current should not notch, and the currents/voltages grow consistently over
> time (until the streamer starts to consume all of the energy in the
> system).  For my "transient" systems i find tuning to the lower pole
> frequency works well because streamers tend to detune in that direction
> anyway, which seems to make the system happy.  Tuning the primary really
> low
> can allow you to effectively increase the "bang" energy to really large
> amounts, allowing you to make really long sparks, provided your silicon and
> capacitors can stand it.
>
> Steve
> _______________________________________________
> Tesla mailing list
> Tesla@xxxxxxxxxx
> http://www.pupman.com/mailman/listinfo/tesla
>
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla