[TCML] Tesla Gun

[Steve Ward <steve.ward@xxxxxxxxx>]

Hi Guys,

Recently got a project running here thats been a lot of fun.  Well, as
fun as fiddling with electronics stuff gets... I think this one
certainly pushes the state of the art in terms of tesla coils driven
by electronics!  I'll cut to the chase with a link to some pictures of
the device:

http://www.flickr.com/photos/kickermagnet/

Where to begin...

Essentially its a DRSSTC that runs extra long pulses to grow sparks at
a low voltage of about 55kV pk.  I believe the reason the sparks stay
pretty straight instead of splitting is that the low voltage allows
growth only with negative coronas, which have been shown to make
branchless sparks, whereas positive coronas tend to branch.  But this
is merely my speculation!  I've not seen any research on spark
development under these conditions (~400khz).

In order to maximally control the straightness of the sparks, the
electronics driving the primary circuit are fairly sophisticated.
Essentially im using a "phase shifted  full bridge", where 1 leg of
the AC output is perfectly in phase with the primary current (so it
experiences wonderful zero current switching), but the other leg of
the AC output is phase shifted, which essentially controls the drive
voltage.  If the 2 AC legs are 0 degrees phase, then the effective
output voltage (differential mode) is zero.  If the 2 AC legs are 180
degrees, then the entire 395V is applied to the primary circuit.  So
what happens is the phase is "ramped" over a period of about 17mS in
order to gradually drive more and more power into the coil.  This
allows for growing long straight sparks that do not branch.  If the
voltage (or phase) is ramped too quickly, the sparks become very
forked and jagged, forming multiple branches in 1 shot.   This
parameter is of course tweakable during operation ;-).

The bridges themselves utilize some cheap little IGBTs (FGH60N60SMD).
Each bridge has been tested to 60A peak, hard switched (due to the
phase shifted bridge control) for 17mS pulse durations at 400khz.  The
calculated thermal rise for the die is 52*C during such a pulse.
Luckily real operation is not this harsh.

The output of 2 bridges (operating synchronously) is combined through
a pair of transformers, with the secondary windings in series.  Each
transformer is wound 6:3, so with 2 combined, its a 1:1 ratio.  The
reason for doing this is 1) i can ground one end of the primary now,
and 2) it absolutely forces equal current sharing between the 2
H-bridges.

Power is derived from a 22.2V 5000mAHr LiPo battery, the type used for
RC helicopters.  It only lasts for about 8 minutes of continuous fun
as the DC-DC converter uses about 700W.  The DC-DC conversion itself
is a resonant type converter used very commonly for capacitor charging
supplies, as its short-circuit tolerant.  The DC bus for the bridges
is then 395VDC with 20,000uF of storage capacity.  One spark eats up
200-250 Joules of this capacitors energy, so even 20k uF is a bit
insufficient, the DC voltage drops by about 30V at the end of the 17mS
burst.

Another interesting feature of this system is a secondary MMC.  That
is, there is a string of 72 x 1.2nF 700VAC capacitors INSIDE of the
secondary coil.  This capacitor adds about 17pF to the coil.  There
are a few reasons for this.  1) due to the gigantic streamers this
thing produces, detuning is a serious issue, so the coil needs to have
a pretty decent amount of self capacitance.  2) more Csec drops the
impedance of the coil, which seems to improve spark growth behavior.
3) adding more energy storage to the secondary coil means slightly
less circulating current in the primary.  The way i think of it is
that the system Q probably stays about the same, but since there is
more energy in the secondary, there is less stored in the primary.
This capacitor also provides convenient measurement of the toroid
voltage by measuring the current through this capacitor only.  This is
where i obtained the 55kV number, which is in good agreement with
previous direct HV measurement of my last QCW system using a Jennings
HV vacuum cap divider.  So indeed, the voltage is low.

The primary coil is wound with home-made Litz wire.  4 bundles of 4
strands of 24awg were twisted such that each strand sees the outside
equally.  Its important to do this, otherwise strands that remain
burried inside the bundle will not carry their share of current.  This
is more surface area than 1/4" copper tubing at a fraction of the size
and weight.  Only needed about 18 feet of the litz, so it wasnt a big
deal to make it using a drill with a jig.

Grounding for the system is questionable!  To be totally un-tethered,
i have a pair of shoes with steel mesh on the bottom for contact with
the floor (dont stand on anything flammable!).  My body is charged to
a few kV with the 2.5A of secondary ground current.  Generally,
though, i prefer to add an additional connection to mains ground.  The
current into mains ground is about 1.5A, meaning my body capacitance
is still seeing about 1A in this condition.  I'm uncertain of the
effects of this current through my body (it must primarily go through
my right arm, which is gripping the conductive gun grip which is the
only connection between me and the electronics).  Anyone have any
research on RF currents in the human body?  People expose themselves
to pretty similar fields when playing with SSTCs and VTTCs, just being
up close to the secondary coil must induce quite a bit of current in
them (according to the mutual capacitance between the person and the
TC).

Well, im sure i forgot to describe certain aspects of the project, so
ask questions if you like.

Steve Ward
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla