I found another way than a transformer, I mean with several inductances. The electric method with a RL circuit. But I need 3 voltages sources or an arbitrary source, I used 3 voltages sources to show it is only simple functions in serial. The source must recover a power so it is not so easy to build but at least easy to test. It is for now a transcient recover of power but not so small. I need only to have a current at start in the coil, a lot, sure, compared to the energy I recovered. Exactly the same than the mechanical device, it is the same functionnality but in the electric field.

The device in NGSPICE is attached. It is a RLC circuit in serial. That's all. The essential function is the DAMPING, it is an exponential function that reduces the amplitude, I don't need to reduce the voltage, but I need to change the slope of the signal because the slope is what I tried to do with the transformer because in my mechanical device I have an angle of application for the spring, it changes the shapes of the voltages sources, exactly like I wanted to do with the transformer, but a transformer is difficult to test because it is in 3D and difficult to have the good shape. Here the damping function helps, it is not the only solutions, like the mechanical device there are a lot.

But now, maybe if people want to test and give feedback. Spice3 dated for a long time now and had be tested a lot since 1983. For the energy the intensity at start is given by Ngspice and the energy at start and at the end of the study inside the coil must be counted. The power dissipated in the resistor must be counted too.

It doesn't work if the inductance has no current inside at start, it is like no rotation of the disk at start in the mechanical device. The voltage source is the spring. The damping function the angles.

Like the mechanical device I didn't find the best angles nor here the best function of voltages.

The last designed with the signals. Not sure about the orientation of L2 and L3, the efficiency must be link with the angles relatively to the horizontal.

V1 is the source of power that rotates the bolt around A1
L1 (red and green) is the inertia of the bolt, rotation around A1 but no rotation around itself.
L2 is an end of the spring
L3 is the other end of the spring
V2 is the spring

V1 is a sinusoide it is equivalent to rotate the bolt around A1 in one direction and after in the other direction
V2 is a sinusoide, when I accelerate the bolt around A1 I need to attract: I use sin(2x), but when I decelerate it I need a spring that pushes, so I use: -sin(2x)

I need to move the spring along the longitudinal axis of the bolt so for that I think I need a AM signal, on V1 or V2. The movement like I study in the mechanical device is bigger on the longitudinal axis than of the relative rotation of the bolt relatively to the support (the support that rotates around A1), so I think the AM modulation is on V1 and V2 is a sinus function but maybe it is a FM modultation. But the signal must be in phase. I mean if the sin(2x) is not well adjust the mean of the energy will be 0, not in one quarter of a period but it could be difficult to measure the energy of the inductances in a quarter of a period because a lot of energy are inside the inductances. In fact the spring must oscillate because the bolt rotates in one direction and after in the other. So the AM modulation must ... oscillates maybe with a modulation greater than 1.

The geometry help to have the inertia around A1 and no rotation around itself. Like I said around itself, it is around the longitudinal axis of the bolt. I didn't put a secondary a charge, I suppose the inductances of the primary have a resistance, so an energy from heating exists, it is the charge.

If the modultation is FM (image of V1 and V2), V1 is FM and V2. V2 must be in phase with V1 I mean in a quarter of a period it is sin(2x) and in the following quarter it is -sin(2x).

It is possible to oscillate L3 + L2 with piezo electric motor but I think the FM modultation must do that because the AM modulation change the force, I need to accelerate the spring, and for the signal if the increase the frequency. But if I increase the frequency of V1 and V2 I don't think it could work. So maybe the fact that have V1= sin(x) and V2=sin(2x) changes the frequency too, in the same time I change the direction of the force.

I drew the transformer with a monophase but it is possible to have more than one phase. It is better to optimize the mass of the iron. All I explained about the transformer can use all technology of the standard transformers: core with iron or ferrite, conductors in copper or aluminium or any other conductor, etc. It is the same technology than the others transformers about the materials, the only particular things are:

- the geometry,
- I use the principle of the méchanical device I explained in another thread,
- the particular shapes of the signals, and the synchronisation of the primaries,
- maybe I need a piezoelectric motor synchronized with the primaries to vibrate the inductances L2 and L3

I used sinusoidal signals but it is possible to use any others shapes.

The helix is needed even in a core.