Universe Breaking Simulation [Not Quite, but very interesting]

So after much simulation I did some experimentation to see how well this idea performs in real life. Here is my circuit:

Zanzal's Flyback LED Driver

I chose not to use a high Farad capacitor and opted instead simply to study it from the flyback. This did provide some interesting data and personal experience I can use in future experiments. Here was the basic setup running in a steady state at 75% duty cycle 14kHz using my 10-tapped transformer core. The exactly transformer specifications aren't too important as the effect will change a little bit but as long as you have a decent amount of inductance and the windings are low resistance it should work fine. My cores are amorphous core, not sure how well it works with other cores, but the effect should be there if not stronger or weaker. For this experiment the ratio of my high side of my transformer to low side is 9:1. This experiment does work well at any ratio, including one with no center tap at all, but the behavior and wave forms is a little different. Its difficult to say how much it mattered, but I really liked the waveform generated by the 9:1 version. Its very dramatic and abrupt where the one with just a single inductor has an interesting characteristic of large ringing on switch closing.

The basic idea behind all of the above simulations is that energy can build up in an inductor over time and that energy can be recycled through a load and finally dumped into a capacitor. Then at 1V your 500F capacitor can be boosted up to 12V and pumped back into your source and recycled again further increasing the efficiency. Of course it takes a lot of input energy to get it to that point and is not entirely risk free to mess with that so I skipped that part and just focused on testing the energy circulation as I was curious how well it performed. Source is a 12V battery. There are 8 LEDs, I think they are 0.5W 2.62V. Input power is around 1 watt.

Here is the wave form showing steady state operation:

Just ignore my crappy scope, I think its seeing ringing and misreading the frequency.

So finally if you read this far now you get to see the effect of how energy can build up over time. Although it was hard to show at the frequency I was operating at above, I managed to tweak the frequency and use pulse bursts to illustrate how over time the energy builds up:

Notice how even a short lapse resets much of the progress gained. This is very important characteristic of the effect shown in the simulator. Overtime small pushes create increasing current flow in the inductor. The simulator while significantly different than real life does do an adequate job of modeling the effect. There are downsides. The increase in energy adds a loading effect. Not terribly bad if you plan on capturing all the energy in a capacitor anyway, but it is a consideration that could potentially limit its usefulness at least as a load on the primary. That effect could prove more useful in future designs that exploit partnered output coils.

Hope you learned something useful, I know I did. I think the effect is worth taking a couple days to study. I would recommend not becoming terribly obsessed with trying to prove its exact efficiency. Even though the effect does scale well when LEDs are used in parallel and all energy can be saved in a high farad capacitor. For me this is something to exploit in the future more than simply trying to build high efficiency lanterns.