Self-resonant solid state tesla coils

advantages:
- always in tune, automatically, because the TC secondary controls the driver. At last, no more manual tuning!
- plug in any TC secondary, runs instantly, no tedious "aw shucks, what's the resonant frequency TODAY" tuning
- this is the most efficient way of driving a TC: transistor switching losses are minimized (zero current, zero voltage switching), which means minimum MOSFET or IGBT transistor heating
- inherently very fast tracking of the resonant frequency and its' changes - no delays
- very simple to build, no expensive "special" parts and no complex circuits like with PWMs, PLLs, or uCs
 
disadvantage:
fast output power modulation, for example with audio, must be done in a pre-regulator (buck step-down, or audio controlled light dimmer connected in front of the SSTC)

In standard resonant switch mode power supplies (SMPS) the resonant frequency (or frequencies) remain constant, and the control circuit has to be tuned just once.

Unlike in these resonant SMPS's, the Tesla coil secondary will change its resonant frequency when the coil surroundings change. For example, moving your hand closer to a small TC coil will detune it: input power draw drops, and streamers get shorter. Especially smaller Tesla coils are easily detuned. The driver would have to be adjusted constantly - either manually, or with some crude way of output voltage feedback. If you already have an existing SSTC driver you could also tweak it to use a simple feedback scheme - for example like on Richie Burnetts SSTC page

The SSTC driver circuit shown here uses a small piece of wire ("antenna") to pick up a feedback signal directly from the secondary electric field. The drive signal is generated by the TC secondary itself, so the driver will never be out of tune.
This simple circuit also includes a 555 based interruptor and you can add a neat LED (a blue one looks cool) for flashy visual interruptor control.

The parts required for the driver circuit are LM555, CD4001, CD4046, two high current mosfet driver ICs (maybe TC4429) or four bipolars like 2n2222 NPN & 2n2907 PNP, and some passive components.

Update note: the old CMOS 4000 series digital logic is slow and has some internal propagation delay. If you want your SSTC to run reliably at much higher than 200kHz, you should replace: CD4046=>74HC4046, CD4001=>74HC02, NE555=>TLC555, and add a 78L05 +5V regulator to supply these chips (the don't run from 12/15V). The DRIVE_A and DRIVE_B signals will be 5V=logic_HI, which still works directly (without logic level translation) for most gate driver chips.

The circuit works quite happily - I've had a small table top demo full-bridge running with this particular driver circuit for 6+ hours continuosly, driving a fist-sized full-bridge. Spectators can draw arcs, touch streamers, light fluorescent tubes, etc, without the operator having to retune the coil. Mosfets and freewheeling diodes don't heat up much at all because the circuit drives precisely at f_res so switching losses are minimized. RF interference caused by the driver board and streamers is really low. In addition, when the TC secondary base is properly RF grounded, the coil won't jump to any other resonant modes than: 1/4-wave with streamers, loaded 1/2-wave when drawing arcs/flames, ground strikes.



Circuit description:

The interruptor uses a 555 timer. One potentiometer sets the on-time, another the off-time. With the component values shown, on- and off-times range between a few milliseconds and some ten milliseconds. I'd recommend connecting a blue LED (D5) for a bonus "wow cool"-effect.

The feedback section around the CD4046 squares up the incoming feedback signal. The signal can be picked up with a short piece of wire pointing towards the base of the TC secondary. (the CD4046 is a phase-locked-loop/PLL IC, but the PLL isn't used here - only the sensitive input buffer section of the chip).

Instead of the CD4046 you may also use a high-speed comparator (high-speed meaning faster than 80ns, t.ex. LM319 or NE521). For an example circuit, see hvguy's excellent feedback SSTC pages.

The RC filter high pass corner freq f-3dB=1/(2*pi*(0.5*R3)*C2)=1/(pi*R3*C2) [assuming R3==R4] should be chosen to be at least four times smaller than the fres of your TC secondary. With 180k and 47nF, f-3dB is as low as 38 Hz, well below f_res, and quite useable for antenna style feedback.
But, much less than 47nF should be used when taking the feedback from a series capacitor on the TC secondary base ground lead - a smaller coupling cap means a higher impedance in the feedback signal path, which in turn prevents too high signal current (=IC destruction) into the CD4046.

The drive logic section buffers and inverts the CD4046 output (DRIVE), and combines the feedback and interrupt (INT) signals. When INT is logic high, the two drive outputs (A=drv-out-1, B=drv-out-2) are set low:

inputs:   outputs:
INT DRIVE DRIVE B DRIVE A
0 0 1 0
0 1 0 1
1 0 0 0
1 1 0 0

When the interrupt signal returns to logic low, one of the drive outputs will change to high, which "retriggers" the SSTC. This helps to instantly start up the circuit again, should the secondary coil already have stopped oscillating (which means the feedback signal will be missing).

The gate drive section of this circuit uses a small gate drive transformer with four outputs, one for each mosfet in the full-bridge. Two TC4429 driver chips (6A peak out guaranteed) drive the gate transformer (transformer design). For this and other driver chips (even those claiming to be latch-up-proof), and bipolar totem pole transistor drives, two 1A 20V schottky diodes should be placed accross the chip / totem pole output: from ground to the out-pin, and out-pin to +12V. This prevents the chip from latching up and frying.


The circuit isn't critical about RF PCB layout. It can be assembled on a proto board or perf board without problems. There should be small ~47uF or similar electrolytic capacitors for the 555, CD4046, and CD4001 on the board, connected close to each. For the TC4429's one >47uF 16V tantalum capacitor next to each chip.


As there may be DC or low frequency content on the half/fullbridge output (due to the drive and interruptor signals), you must use a DC blocking capacitor in series with the TC primary coil. That'd be just a normal pulse foil capacitor (Wima FKP1, Wima MKP, many others) that is rated for the mains voltage and has >220nF.


(todo: add some operation pics)
See some preliminary pics very low power as safety precaution