[Prev][Next][Index][Thread]

Re: Magnetic pulse Tesla Coil needs megawatt directional co (fwd)



---------- Forwarded message ----------
Date: Mon, 30 Mar 1998 10:38:36 +1200
From: Malcolm Watts <MALCOLM-at-directorate.wnp.ac.nz>
To: Tesla List <tesla-at-pupman-dot-com>
Subject: Re: Magnetic pulse Tesla Coil needs megawatt directional co

Hi Scott,
           I printed this post off the HV List and carefully read it 
during the weekend. I have a few comments, the first of which that it 
is a little difficult to tell exactly where your 10mH inductor fits 
into the finished item (apart from being an energy store).

> From:  Scott Stephens [SMTP:stephens-at-enteract-dot-com]
> Sent:  Thursday, March 26, 1998 7:52 PM
> To:  tesla-at-pupman-dot-com
> Subject:  Magnetic pulse Tesla Coil needs megawatt directional coupler
> 
> I have been researching methods and striving to build a compact, high power
> and even solid state tesla coil. So far I've discovered numerous reasons why
> this is not a trivial task.
> 
> I hope to avoid re-inventing the wheel, loose money, damage myself and my
> environment with hot plasma, high voltage, ionizing and non-ionizing
> destructive radiation, chemical pollution and violent neighbors enraged over
> another lab accident, (so far I've survived all the above).
> 
> Capacitive discharge TC's may store 1 to 20 Joules of energy per pulse or
> discharge. Since E=.5CV^2, a typical cap might be 50nF at 20KV
> (10J=.5(5*10^-8F)(20,000V)^2). A 50nF 20KV poly-oil cap could require 30mil
> poly, of area C(nF) = .013A(sq. inch) (derived from
> Cpf=.223EeffArea/thknessMils) so the cap needs around 5000^2", which is a
> roll around 1 cubic foot (a roll 10" high and 2" thick with 20 layers of
> 50mil/layer -at- 10" avg. diameter). Filled with oil, it will probably weigh
> 30lbs. A big, heavy, ugly, gooy mess.

Interestingly enough, this mirrors exactly my reasoning in choosing 
to buy the two CP caps. A good way to get 20J.

> A FET switches a 200 volts across an inductor of 10mH for 2.25 milliseconds
> (ms); 200V for 2.25ms will charge up to (200V=10mH di/dt) or 45A, and store
> E=.5LI^2=10J. This 10mH inductor will take (if I'm guesstimating correctly)
> a slug of volume around 2 cubic inches. Far more compact than the cap.

The slug is quite incapable of storing that amoung of energy. Most of 
the energy is stored in the air around the slug. The gap is so large 
that you would be better off with no core (and non-linearities) at 
all IMHO. That amount of metal isn't going to vastly improve the 
induction factor.

> But how do you get the energy out in a form that can 1) drive a tesla
> primary and 2) not destroy a wimpy FET?

I am not at all clear on the total setup. Could you post a simple 
ASCII diagram please? Forget jpgs unless they are uuencoded. My 
mailer can't handle the various arcane formats that come my way. 

> The FET's ON time is not too significant, since the inductor will naturaly
> take time to charge up - its rate of change (dV/dt) is limited by its
> inductance. But the FET's OFF time determines the rate energy will flow out
> of the inductor. I want to open the FET in 5us (microseconds) so it will
> drive a 100KHz TC 2ndary.

FETs can do an awful lot better than 5uS. Output voltage is limited 
by all capacitive loading on the winding assuming resistance is 
pretty low. You can easily quantify this: V = I x SQRT(L/C)  Note also
that the time that energy transfer takes from L to C is governed by 
the resonant frequency of these components: it takes 1/4 cycle to 
transfer the energy so Transfer time = PI x SQRT(LC)/2

> Now the inductor absorbed 10J in 2.25ms resulting in around 5000 Watts
> power. It will release 10J in 5us or 2 megawatts of peak power. After 5us of
> the FET open, the inductor which 5us prior had 200V across and 45A through
> now has (V=.01H 45A/5us=90,000V) or 90KV! Poor FET went POOF! Stopping that
> current was like trying to stop a freight train.

The classic solution is to use two windings sharing the same "core" 
(nee a flyback). Suppose your supply is 200V and you want 20kV out.
Turns ratio = 100/1  (sec/prim)  Now the FET has to withstand at least
2x Vsupply assuming output voltage is limited.

> One solution would be to place a spark gap on the inductor, and let the gap
> fire to deliver its charge to a TC primary. All the energy would need to be
> absorbed in the first cycle, as the next oscillation a nasty voltage level
> would appear at the FET drain (FET is connected common source). So a high K
> value, like .999 would be necessary with a hot spark mandator.
> 
> Along with mega-watt TVS or transorbs on the FET. They eat power. Yuk.
> 
> What is realy needed is a directional coupler. A directional coupler between
> the FET-Inductor and TC primary-load would allow energy to flow from the
> inductor to the load, and redirect any unabsorbed energy from the TC load to
> a resistor. With a K of .5 such as with a magnifier coil arangment (the
> switched inductor would serve as the magnifier coil) maybe 1/3 of the energy
> would make it to the 2ndary.

You can do an awful lot better than 1/3. A HV diode is typically used 
as the steering element in flyback designs.

> I know how Tesla coils are built. And microstrip hybrid directional
> couplers, but how do I:
> 
> 1) make a directional coupler to tranfer energy from a flyback inductor, to
> bottom-feed a Tesla Coil? Guanella hybrid transformer? What material &
> construction issues happen with kilo-amp currents?
> 
> 2) make a coupling network between the flyback inductor and directional
> couple that rejects/reflects 2.5ms charging voltage/field and transmits 5us
> flyback pulse? 'Simple' LC network?
> 
> 3) (extra credit) Save or conserve any reflected energy from the Tesla Coil,
> via the reflected port on the directional coupler. Another directional
> coupler? A circulator?

Try an energy recovery winding. An additional benefit is that it can 
be used to clamp Vds to a specific voltage above Vsupply if the 
seconday ever becomes unloaded. 

> Surely this has been done before. I've seen a neat circuit called a magnetic
> pulse compressor/magnetic switch. Uses inductor saturation as a switch for
> high power. A good example of one is at Allied Signal's web page, the one
> that features that ass-kicking Metglas core material.

Perhaps I'm misreading what you are trying to do. A diagram would be 
ace.

Thanks,
Malcolm