The driving circuit for the HV TV flyback transformer is very much in-topic of this thread right now.
If you are willing to change your approach a little and forsake using your signal generator+amplifier to drive the primary winding of the flyback transformer, then you may try this simple efficient high power proven circuit.

If you use a variable capacitor for C1 then you can vary the oscillation frequency continuously, just like with a signal generator.

I have published schematics of Tigers device - all what "group of Wesly": says is quite  different because they are like you - they are "in progress of trying to replicate" device of other creator!
PS  http://germaschka2011.narod2.ru/

That is the principle of the coil Tariel:
ball-passing pulse with a flat coil
Feeder-ball "primary" coil in the video "Aquarium", she is a high-frequency magnetic field - the transmitted pulse accelerator
In a coil through which a pulse will be induced current

No ferrites in the coil can not be Tariel, a brake pulse!
http://i081.radikal.ru/1111/82/53ef422cc7b7.jpg

  inertia up and inertia down  but first we must learn how to produce this inertia wave 

Basically anyone who has bought or is using a standard sig gen MUST use some kind of an amplifier like this to pump at least 10 watts into the coils.     The blurb from the website about this topology says: "  The main difference of the half-bridge converter...

...and for non-rectangular waveforms this must be a high-power linear analog amplifier.
The simple Half-Brifge two transistor circuit does not work as a linear analog amplifier !!!
It is a double Class B amplifier that's good only for rectangular waveforms.

For non-rectangular waveforms a Class-A or Class-AB amplifier is needed.  Linearizing the H-Bridge is such a chore that it's better and cheaper to buy a premade amplifier.
There are many Video Output Amplifier or Line Driver integrated circuits that perform this function well from DC to 100s of MHz at power levels up to 25W.

The STK392 video deflection amplifier that the STAAAR team has been looking into at maxes out at 1MHz.
There are faster and more powerful Video Output Amplifiers and Line Driver ICs out there, ready to be used (made by e.g.: Sony, Phillips, MSK, AnalogDevices, etc...)

Interesting experimenting with this stuff ...

It reminded me again of Otto and his dedicated efforts before he passed on.

I repost his doc file here, have a read and see how similar his work was

Do someone has idea how to measure possible neutrons radiation from device ?

Severall answers:


@ verpies

1 Ohm measurements of my FG's: http://www.youtube.com/watch?v=CJQt1mD7bIc

100 Khz / 50% duty

Philips  Sine   5.4  V
         Square 8.8  V

China    Sine   2    V
         Square 1.92 V

 
@ All

caution about using a transformer as 50 Hz input as they will burn out.
Use a light bulb / resistor in series to protect the transformer:
http://www.youtube.com/watch?v=Izyg4TAh8jQ


Replication attempt of Aidas 100 Watt RF output phenomenon (unsuccessful):
http://www.youtube.com/watch?v=SXOUd9N5R3g


@ Wattsup

Minimum effect on 150 Khz,  but some 105 V PP output on 276 Khz, so
i see similar (less) results, but on an other frequency (not surprisingly):
http://www.youtube.com/watch?v=dO-6l6ifWwo

Regards Itsu

@itsu

I think your signal generators are of decent quality and they do not introduce severe spectral artifacts even when driving low impedance loads.
I was surprised that the Chinese SG had less sine and square wave distortion across 1ohm load than the Phillips SG.

There is still a question how your signal generators react when they are forced to sink current, e.g. when an inductor discharges some previously stored electrical energy into them.
Other than that, your sinewave sweeps of the various windings should yield worthy data.

Also, I think that the light bulb in series with the ~50Hz 9VAC transformer and 15t winding is an excellent idea !
If this light bulb dims on that 15t winding at ~50Hz as you are varying other parameters of the system, it will be the first sign that something unusual is happening. So keep it there while doing all the future tests.
Also it's better to use light bulbs with a straight filament rather than a coiled filament in order to avoid introducing unwanted inductances into the system. 

Increasing the output voltage of the above ~50Hz transformer will not hurt either. 
However, if you want to have the ability to vary the LF ~50Hz frequency a little (just like Aidas recommends to harmonize with the HF oscillator), then you can drive the 15t winding from an analog automotive audio amplifier driven by your weaker sine wave signal generator.  Those car amplifiers are dirt cheap and designed to drive low impedance speakers.

This car amplifier trick will not work for the HF though (but a video output amplifier would).

I am too overwhelmed with data at the moment to comment on your spectrograms in detail. So before I can analyze them, I am just sticking to observations that I am very sure about.
One of them is that Wattsup's & Aidas' magic frequencies should not work in your device.  You have measured different frequency peaks in your experiments....

P.S.
Could you add static captions at the bottom of your future videos that state where each of the 4 windings is connected? e.g.:
t1 - NC, t50 -> Ch1, t15<-50Hz 9VAC, t150->50ohm
etc...

Not that I need these captions in your videos (your English is 100% clear to me), but I am sure that there are other viewers that understand only 5% of what you are saying.
However most importantly, I would like you to set an example to the STAAAR team how to properly document their tuning procedures.

@ verpies,

thanks for the comments, they are much appreciated.
I will look into those audio/video amplifiers.
Presently waiting on some variable capacitors which makes tuning easier.

I will keep in mind what you are saying about subtitles  :-)

Regards Itsu