The reason why L2 is magnetically coupled to L3 & L1 is not magnetically coupled to L3 is because L1 is aluminum wool stuffed inside of the copper spheres and the copper tubing which connects the spheres with each other reflecting the inductance of L2 with reversed polarity (just like a mirror) from inside the copper tubing. The paramagnetism of aluminum severe ly reduces the direct coupling of L1 to L3 towards negligible values. L2 has 60k Ohms of series resistance and is wrapped around the copper tubing. L1 has a series resistance of 2k Ohms inside of the copper tubing serving as its electrode. The aluminum is also acting as a self-referencing (parallel) capacitance internalized inside of the copper tubing. Tantalum may substitute for aluminum? This capacitance is simulated with the help of 1 Farad, each, of parallel capacitance placed inside of the simulated inductors, L1 & L2. L3 is a motor load of 25 AWG copper winding possessing 10 Ohms of series resistance and no parallel capacitance. All capacitors possess 3 Ohms of equivalent series resistance. C2 prohibits the escalation of impedance at L3. The magnetic coupling of L3 is to its armature, not to itself, since the copper winding of L3 contributes a much smaller coupling coefficience than the contribution of its ferromagnetic armature. X1, X2, X3 & X4 are spark gaps filled with air possessing a voltage threshold of one kilo volt. Gear Approximation Method is simulated with RELTOL (relative tolerance) equaling 1.

The frequency of the sine wave generator, V2, is slightly faster than the frequency of the sine wave generator, V1, by 10% to create a beat frequency between them. This helps, or insures (I forget which!) the overunity gain.

The voltage precharged onto capacitor, C1, also regulates the overunity rate of gainful output.