Off-Line Tesla coils (OLTC)

Original poster: "Terry Fritz" <>

Hi All,

I have been doing some studying of off-line Tesla coils.  These are (would
be) Tesla coils that have "no" HV transformers but run directly off line
voltages.  What I found so far (purely by computer modeling) is somewhat
encouraging.  While a far cry from a 38 foot streamer unit, it may easily
approach an NST powered coil in power.  The only real new components would
be an IGBT solid state spark gap and rather odd MMC caps.  Everything else
is quite conventional.  So far, the design would run off 240VAC at around
700 watts.  This is the schematic as it now stands:

The the top is 120-0-120 VAC (single phase split 240 as is common in US
homes).  That voltage is rectified by two diodes and charges two 100uF caps
(50uF primary capacitance) through some ballast resistors.  This gives a
firing voltage at 60 BPS SYNC of:

240 x SQRT(2) x 2 = 679 VDC

The capacitor waveform looks like this:

Since the caps need a full AC cycle to charge, the natural firing rate is
limited to 60 BPS.  The current through the charging diodes looks like this:

Only about 4 amps RMS and very easily in the range of any 240VAC circuit.
One concern is the RF spikes at about 20 amps which would have to be
filtered.  The resonant frequency is 16 kHz.  A sweep of the circuit revels
a dual hump noting that the circuit is rather simplified so they appear
higher Q than in real life.

The current in the primary and secondary are shown in:

The IGBT "spark gap" would have to withstand a 2000+ amp spike 60 times per
second.  I would quench on the first notch electronically to trap the
energy in the secondary.  This first notch quenching also reduced the
stress on the IGBT and caps considerably.  The primary RMS current is near
120 amps(!!) so turning off the primary as soon as possible really helps
losses and stress on the primary caps and IGBT.  The secondary currents are

The primary coil would be a single turn of 1 inch copper pipe.
Surprisingly, the MandK program predicts I can still get very good coupling
with just a big single turn primary at about 0.15.  I thought having only
one primary turn would devastate the coupling but that does not appear to
be the case!  The secondary is at 2.17H (2170mH) which is much large than
usual to accommodate the low Fo frequency.  Aside from having many turns,
it is conventional.  Since the IGBT can switch with very low loss, the vast
majority of the energy is sent to and trapped in the secondary coil.  In
this case, the energy is:

1/2 x 50uF x 679^2 = 11.52 joules

At 60 BPS, the power delivered to the secondary is 11.52 x 60 = 691 watts
which is more than most LTR 15/60 NST systems can claim. 

The models does seem to fall short in one area.  The 16kHz the streamer
load estimate model of 220k+1pF per foot of streamer capacitance seems to
give a very light streamer load...  This is a big issue since the estimate
may be "right" and streamers operating at only 16kHz may not work well!!
However, with 700 watts trapped in the secondary system, one would think it
would go somewhere?  A major theoretical hole exists here that at least "i"
can't predict...  The model gives a very lightly loaded coil output that I
really do not trust for accuracy:

So it seems to work in the computer ;-)  No guarantee that it will "really"
work, but it has passed the significant hurtle of working in computer
models.  The computer also eliminated a number of other designs in an
afternoon that it would have taken months (and thousands of $$$) to
eliminate by testing real hardware...  

Note that the Ls/Lp ratio is 1,000,000:1 so Vs/Vp =1000 or 679000 volts.
This addresses Antonio's concern that the primary circuit must have very
low inductance at such low voltages.  Fortunately, the coupling was not a
great concern so just a "one big turn" primary could work.

The special concerns that computer modeling predicts are as follows:

1.	The AC line is poorly isolated from the Fo frequency in the primary caps
as the circuit now stands.  The 16kHz can easily push high currents back
into the line.  One solution may be to replace the didoes with FETs that
would switch off during firing.  Simple filters may take care of it too.
The charging circuit works, but it could use much refinement.

2.	The primary capacitors would have to be 700VDC 120Arms 2500Apeak at
100uF.  Since it is a voltage doubler, we have about 4X the amount of
capacitor.  Capacitor arrays can easily get to these specs but they get
costly at about $300.

3.	The primary resistance would have to be very low at about <0.05 ohm at
16kHz.  With a big single turn primary "pipe" and a very close IGBT and
cap, this seems very possible.  Many thanks to Gary Lau for his
excellent(!!) work on this at:

5.	The primary currents are "high" suggesting a 2000 amp peak 1000V IGBT
and a cap RMS rating of 120 amps.  However, this seems to be well within
the reach of existing devices.  10 Digikey IGBTs would do that and in
quantity their price is very good.

4.	Depending on what the streamer loading really turns out to be, the
coupling may have to be adjusted for maximum power transfer.  This should
be simply accomplished by raising the secondary in any case.

Special notes about such a coil are:

1.	Being an IGBT spark gap coil, the gap is very low loss.  The only sound
it makes is the high currents softly "clicking" in the circuit from
magnetic constriction.  But the streamers should be very loud especially at
16kHz ;-)  

2.	With such a low AC input current, conventional (cheap) small fuses can
be used allowing very fast fusing action if something goes wrong.  

3.	The low 16kHz is in the audio range.  The low Fo frequency and the IGBT
drive may vastly reduce radio interference.  However, the low Fo frequency
may make the output streamers considerably more dangerous.  I am not sure
how nerves react at 16kHz?  Also the human body's impedance at this
frequency may look like a very nice "load".  Of course, there is no skin
effect in any case. 

4.	No transformer, spark gap motor, or variac.  100+pounds lighter ;-))
The electronics can easily be run from a few 9 volt batteries.  A truely
"different" coil...

The questions I have for those that may know:

1.	Does anyone have any data on Tesla coils running at near 16kHz and their
streamers?  Do streamer still form ok?

2.	Does the human body "feel" 16kHz (nice sine wave) high voltage signals?
Not that I am planning on finding out the hard way ;-)

3.	Does anyone know of a reliable source for say 1000V 2500Apeak IGBTs with
built in reverse diode.  Quantity one or two...  Affordable...  IGBT arrays
may be the way to go here too...  MMIGBT (sorry ;-))

4.	Any ideas on a better charging circuit?  Note that the diodes are
conducting during the firing of the gap at the negative part of the cycle
and only the 20 ohms of resistance provides isolation (but pretty good). 

If all seems well I will try to make it.  The caps and IGBTs will cost real
money but not too bad.  Since it is low voltage, I have a much better
chance of not blowing up a lot of IGBTs.  Of course, those big ones don't
blow up easy...

Hope this is of interest and at least amusing ;-))