Construction Details...
L1 is the "inner", primary, coil of the Ammann brothers' Atmospheric Generator's transformer. To represent its parallel capacitance (of one Farad), it is filled with aluminum wool which has been "conditioned" by first using this wool as the terminus of one electrode while another electrode of some other material are both immersed into an electrolyte of borax (or, baking soda) and both electrodes are energized with an A/C current to cause a layer of alumina (aluminum oxide) to form on top of the surface of the aluminum wool, and then stuffing a copper pipe with this conditioned aluminum wool, and then holding this copper pipe in a vertical orientation, and then positioning a bunsen burner flame underneath one end of this copper pipe to cause a stream of hot air to rise upwards through the interior of this copper pipe to dry out any lingering moisture which may be adhering to the surface of the conditioned aluminum wool.
The Henrys of L2 is defined by its diameter of bare iron, single layer winding (not by its mass as is the case with conventional windings {of today} predicated upon when winding copper - not iron - coils). L2 is wound upon a wicker-style frame of iron rebar setting each turn of winding apart from its previous turn to create a capacitant spacing between each turn of winding.
The coupling coefficient of L2 and L3 is 99.9999999% due, not to proximity between its pair of coils, but -instead- due to their coupling is between their iron masses: the mass of iron armature upon which is wound the stator and starter coils associated with L3, and the iron wicker-winding of L2. In actuality, the distance of their separation is a radius of several miles determined by the radius of L2 (simulated by the inductance of L2).
The inductance of L1 is weak due to it being constructed of mere copper tubing. It's resistance is also low due to its shrunken mass.
The diodes, D1 and D2, have their cathodes pointing towards the parallel capacitance of the aluminum oxidized wool of L1 which, at first, may seem backwards since the oxide only forms if the aluminum is serving as the anode of a diode. But the cross-wired capacitors, C5 and C6, are effectively reconnecting the anodes of these two diodes with their origin at the parallel capacitance possessed by inductor L1 by the use of a cross-wired connection. This cross-winding lends stability to what would otherwise be considered an unstable circuit (which is what conventional wisdom calls overunity, free energy circuits: unstable). This makes the task of Micro-Cap simulator a bit easier to compute the outcome with less error which tells me that this is also a good idea to use this feature in the actual, physical build.
The circular polarity of L1 relative to L2 is significant. They must possess a continuously, unbroken orientation of direction to their windings.
The slight difference of frequency between sine wave generator, V1 and V2, is intended to create a beat frequency. It's use of a very low frequency of 1mHz yields greater amplification than would higher frequencies. Its low voltage of 1 micro volt input also works in its favor by not suppressing the germinal reactance which must be fostered, and not obliterated, during its initial stages of development by avoiding the use of a large voltage input.
The precharged condition of capacitor, C1, is one factor which regulates the rate of amplification of output. A greater voltage input, here, accelerates the growth of output.