I think the duty cycle should be changed in a much wider range, say from 5% to 50% at least. If you cannot do this
with your present PWM generator, then just build a simple 555 timer circuit like shown herehttp://overunity.com/8411/steorn-demo-live-stream-in-dublin-december-15th-10-am/msg243175/#msg243175
The frequency and the duty cycle can be varied independently from each other.
I think somehow you would need to make sure (as a first step) that the input core should go into saturation.
To do this, remove the permanent magnet(s) from your full setup and insert a 1 Ohm (maybe 0.1 Ohm) noninductive resistor
in series with the two input coils to monitor input current by your scope. Here is a good video on this where
he changes duty cycle and first the input current of the coil increases linearly as it should, then there comes a break
and current starts increasing non-linearly: https://www.youtube.com/watch?v=oDt_Im1TtEE
He speaks Greek language, on his blog there is the schematic for his measuring setup: http://tinyurl.com/jmrxrxs
The goal would be to find a low duty cycle (the lower the better) setting with a given (variable) supply voltage amplitude and
input current value where the core goes well into the saturated area.
And when this is more or less achieved, and still there is no output power across the load of course, you could insert
a permanent magnet. I believe your present magnet is very, very strong. I know you tried ceramic and AlNiCo ones too
but now I suggest to monitor the input current when you insert a magnet to see how it influences input core saturation.
I think there would be a certain (unavoidable) interaction which could be minimized by using the lowest duty cycle possible
for a good input core saturation with a given input voltage level.
Do you have an air gap under or at the output coil core? And I assume your input cores are two C shapes facing each other,
without air gap, right? Or you use an E core there?
On tuning the output coil with a capacitor: probably the 100 Ohm load attenuates the LC circuit too much to see the usual voltage peak
at resonance. First try to use some kOhm load (or even no load) to easily find the resonant frequency either with changing capacitors
and / or sweeping the input frequency.
No need to place a capacitor in parallel with the input coil.
I suggest reading member Getca posts in his findings, he reported COP>1 result. http://overunity.com/4300/a-truly-overunity-transformer-meg/msg134162/#msg134162
I attached a video, taken from youtube long time ago, that nicely show how permanent magnets can passively cause local core saturation
to create a virtual "air gap" in the core. Unfortunately I cannot remember who uploaded it to give credit to him.
In your video ( https://www.youtube.com/watch?v=Iz4POclJ8ws
) there is still output power when your permanent magnet is horizontally attached
to the side of the lower ferrite bar and power disappears only when you completely remove that magnet. This may indicate it is a
very strong magnet for the job? Or too small air gap at the output coil side in the core? There exists a strong magnetic coupling between
the input core and the output core in your setup
As another test, you may wish to swap the places of the permanet magnet and the input core as per the animated gif file shows in Getca post.