Re: It works! Solid state driver

From: 	Harri Suomalainen[SMTP:haba-at-cc.hut.fi]
Sent: 	Thursday, December 04, 1997 12:32 PM
To: 	Tesla List
Subject: 	Re: It works! Solid state driver

On Thu, 4 Dec 1997, Tesla List wrote:
> I use to think the MOS input meant no power. How wrong I was! At 100KHz, the

Well, it consumes almost no power. However, the usual way is to
brute-force charge/discharge the gate capasitance. That takes power.
It would be possible to recirculate the gate-capasitance stored energy
for better use but it is hardly ever worth the trouble.

> drivers, when a plasma ignites. If I use end-feeding with a K of maybe .6,
> and a hot spark ignites, and I use a ferrite transformer with a K of near 1,
> then I'm effectively shorting the FET's. They will die before I can pull the
> plug.

Go for K=1. Higher K will mean the energy is transferred to the secundary
side much better. You still need to make your hard-ware able to take even
full short circuits (like saturation of core during transients etc). That
is usually done in almost every SMPSU too. Usual reaction times for such
systems might be around 100ns or so. That is usually quick enough to
protect the switching devices.

> So I'm interested in using a ferrite block between primary and secondary
> windings, to give a more modest coupling for leakage reactance ballast, and
> pulsing the Fets with a low duty - cycle high current impulse, in the hopes
> no high current streamers will stay lit long, and keep the secondary Z high.

As the primary discharge occurs the secundary goes from normal load to
very high impedance. Then the arc will not short-circuit anything.
However, the sudden huge change in load (removal of it) will easily cause
other problems, mainly saturation of transformer.

This effect is not noticed with some discharge electrodes. I've put around
1kW through a prototype having a needle-like electrode (coil upper end
wire just pointing up). No problems with K=1 and no current limiting
circuitry. The same prototype wouldn't have a chance of survival with
toroid discharge electrode. With toroid it will die out at the first
spark due to load-change problem. Gapping to core might help a lot.

> Continous, 8A, but pulsed, up to 40!. Of course, with 1 ohm Rds, efficeincy
> will stink.

Go for maxinum duty cycle. Reducing duty cycle in a cw-device is not
helping you with anything. The coil will be storing the energy pumped in
anyway untill it has enough to strice arc. To get a single cycle strike
you'd need quite an energy pulse I think (like spark gap driven ones).

> And using transient suppressors (zeners) across Drain-Source.

Not needed in a properly made bridge topology system.

> And maybe even external (Hexfred?) diodes to ease the dissipation of the
> MOSFET's internal diodes. The worse the primary tuning (too little C), and

Be sure to include a series diode for fet in that case to make sure which
diode actually carries the current: fet parasitic body diode or the
external one.

> I plan to use a 600W, double-filtered triac dimmer, because I don't have a
> variac.

Triac dimmer will fire when the voltage has risen high enough. Make sure
you are using one conducting at more than 90 degrees phase lag when the
voltage is decreasing after it has fired. That type of voltage limiting
will work. Make sure it can take the continuous surging to a *capasitive*

> And of course, my logic control will be isolated from my FET's & drivers,
> through 15 feet of steel conduit, with isolated grounds on the driver's
> coax, and shielded boxes, to protect it from what must be some VERY ugly EMI!

Oh, c'mon. I used no shield nor any ground planes to protect from emi.
At 1kW the coil being a couple of meters away made no problems with
anything except perhaps some very high-impedance circuits with it did not
have. Scope worked of too.

> One neat trick which may be possible, is to use a sense antenna to pick up
> the secondary voltage, and use it to drive the coil. Basicaly the feedback

Well, just get a scope lead and don't connect it anywhere. It is
high-impedance anough to see the resonant peak. Works great perhaps 1-2m
away from coil with even modest power like 200W or so.

Harri.Suomalainen-at-hut.fi - PGP key available by fingering haba-at-alpha.hut.fi