Re: It works! Solid state driver (fwd)

---------- Forwarded message ----------
Date: Thu, 18 Dec 1997 06:29:31 -0700
From: Scott Stephens <stephens-at-enteract-dot-com>
To: Tesla List <tesla-at-pupman-dot-com>
Subject: Re: It works! Solid state driver

>> Maximum coupling is a great feature of solid state design with end feeding.
>> But if you load the secondary (with a white-hot discharge, for instance) the
>> decreasing load impedance of the secondary will now appear (to the turns
>> ratio ^2) across your FET's :(

>Actually, the base impedance *rises* if the other end of the 
>secondary is loaded. Check a reference on 1/4 wave transformers. 

Yea, that's right! I remember now. And I understand now that the big, ugly
surge happens when the secondary is cold. Like putting a car into motion.

But our coils only have one 'gear'- unless we care to wrap multiple
primaries, and use switching logic to switch into successively lower Z
windings, as the secondary rings up.

This doesn't initialy seem to make sense; the secondary has a fixed
impedance, and the end-coupling transformer does too. But the secondary as
an energy-storage device dynamicaly changes impedance. What operating phase
do you match impedance to? Ring-up low Z, copper-loss loaded or sparking loaded?
All the above, with current regulation.

You don't have these problems with spark-gaps. I'm realy starting to think
about using Fet's & Ferrite for a kilo-watt, low KHz power inverter, and get
the oil and capacitor polyethylene out. At least it would still be smaller &
compact in relation to 60Hz pole-pigs.

Two solutions:

1) Gun up the motor slooowly...Trigger the oscillator to start at 60Hz
zero-crossings, and stop under heavy load, like after a discharge. Line
power rectified to 120Hz, will ramp-up in around 4 milliseconds. A 100KHz,
Q=300 (before it lights) will ring up around the same time. How convenient :)
If the ramp is switched with one 250 volt semi, cheap and common low voltage
BJT's can be used to do the switching.

2) Frequency/pulse width modulate the input, so the drivers "see" a higher
impedance. A PWM regulator will do this. FM high-low chirp would faster, but
with a switching-loss penalty. At 100KHz, the FET's & Ferrite's don't have
much left in them.

>Use a separate current transformer to measure current (if transformer
>coupling is wanted). Make sure CT won't saturate before it is capable of
>tripping the over-current system.

>Oh, forget the poles etc. The coil driven at (almost) resonance will
>appear as pure resistive load. After it has stored enough energy it will
>give a spark (perhaps after a few cycles or so). Then it will go into
>*high-impedance* state. Current protenction comes here if the transformer
>is not gapped enough to take the transient. Then, pure resistive load
>again. By resistive I mean it is like a resistor but draws sinuoidal
>current because of the series LC-network.
>You can use transmission line theory to calculate the power it takes
>(the value of the "resistor"). If theory is not wanted just try it out.
>Make 1st eg. 1:2 transformer turns ratio. Add more turns for higher
>voltage and higher power. Test it every now and then. Finally, it takes
>just the amount of power you want. Easy, uh? No simulation required.