Hi,
Here is my version of Melnichenko coil.
Big aircore primary and the smaller secondary that is on a core.
I used IRFP460 mosfet for switching, the power supply can go to up to 200V.
Currently using 120V and ca 24% pulse width, input is tuned to near 100W and output energy recovery is measured on the load1 that is collected from the primary (CEMF). Best efficiency I reached was 86% without the secondary load connected.
When connecting the second load, the load1 power recovery falls to 74.5W. I haven't measured the secondary power (lot of work), but this lightbulb is actually not so bright (picture shows it was bright). It looks like that the energy that the secondary lightbulb eats up is taken from the primary in a same amount.
Of course - I tested different coil positions as well. Tried to touch the secondary with a ground wire (no effect), of course - I haven't tested the grounding for the primary, because have to be sure that the PSU will tolerate that biasing.
What was interesting that with a secondary load there was some sweet spot of pulse width where lights were the most bright.
Connecting the secondary light does not change the input (or a few watts up)
I haven't cut the core in pieces yet as you can see. It would be interesting to test 2 or 3 cores in series where cores are isolated from each other. This was Melnichenko's recommendation as well.
Output power measurement is done by the scope by multiplying V * I in a real time. Input power is shown on PSU. Values/proportions are consistent over the different power range.
Hmm... this is how it is...
Best,
--Raivo