I can shed some light on this phenomenon for everyone because I know there are doubters among the masses. I have been able to get the same results as LasserHacker not by coping what he did but by understanding what he did. The secret to all of this is the coil dynamics vs the load as I said, when the fall of the field in the coil is at a greater potential than the load there is a little fluke to it in that it will produce a tremendous current rush in the load I know from hundreds of burnt out LED’s I have amassed in my test. There was a strange thing I could not put my hand on for over a year ( no scope to work with ) and discovered when the load is properly matched to the coil it matters not how many LED’s are in parallel but the series voltage as I said the collapse happens much faster than the field build up. This is directly proportional to the voltage supplied. The current is but a function of speed and greater speed equals greater pressure as you will note from a water hose when you decrease the output stream you inversely change the pressure.
By using what we would normally shunt it no more than lost energy recovered. The most important thing is the balance between the two for if the load is too small the current will be too high from the stagnate magnetic field we have left in the core, and too much and we do not get the results we are looking for. Example in the video even though I have three LEDs hooked up they are in parallel not series. I will also note my circuit I use a resistor and capacitor not shown in print but needed some times and that resistor size matters as smaller equals greater current draw and sometimes you would want greater field strength for your application. As I noted mass of the core does make it operate for that is the stored charge ( magnetically ) and just like a capacitor too small it will not work and too big and it is hard to get the magnetic charge you want. You can easily find the right direction with out a scope by watching the current draw. When it is set up properly you can take a 1n4001 diode from the coil hooked to a filter capacitor and supply the LED’s with that at less current than they would normally consume ( they still use the same amount of current but because we have over unity we use less ). In fact I can supply three LED’s from that setup with 2 – 10 uA current draw supply but have 3 – 8 mA on the LED’s that have a greater voltage than the input. So that my friends is how you get something from nothing and there is no magic fields or mirrors just science and math.