Kapanadze Cousin - DALLY FREE ENERGY

[itsu]

Minus the forward voltage drops of the PS diodes.
Yes, but I doubt the MOSFETs are ringing with the inductance of the primary winding, because this inductance is simply too large to yield such high frequency. 
Pls use the formula C = 1/(39.478 * f² * L) to calculate how low of a capacitance you'd need to have, in order to get a 14MHz LC resonance frequency (f) with your primary inductance (L).

Also, a time-domain scopeshot of the drain waveform would be useful to identify the ringing stage.
Pls try to connect a parallel 47pF cap to the gate and/or drain of the MOSFET in order to locate where that LC parasitic is exactly.  Look for downward frequency shift.
It could be related to the placement of the air gap under the windings or the capacitance of the winding but doubtfully related to the inductance of the primary winding which is simply too large to resonate at this frequency with realistic capacitances.
First of all, these additional peaks cannot be related to the yoke just changing its inductance, because such change would only cause a frequency shift of the existing peak.

New peaks cannot appear due to a change of inductance, but they can appear due to any nonlinearities introduced, such as ferrimagnetic saturation^* ...and other effects.
It is not so chaotic:

First of all, the peaks at 18.833MHz, 28.250MHz, 37.6666MHz, 47.08333MHz are all consecutive harmonics of the 9.416MHz peak.
That leaves only the small peak at 35.25MHz as the odd one.

The 14MHz is also unrelated to any other peaks, but notice, that this 14MHz peak does not belong to the group of peaks that was caused by the permanent magnets.


^{* It is unlikely, that these relatively small and weak ceramic magnets could saturate such a large yoke core.}

I measure the both primaries (only the C to B and the F to D part in your loss-less clamp diagram) to be resp. 62.5uH and 64,2uH.
For 14Mhz that comes to about 2pF on your above formula, so that can come from anywhere.


I will try to capture a scopeshot of the drain and some further tests tonight.

Itsu

[itsu]

Exactly! Just to add on your explanation that the diode acts as a switch. If for example a diode is connected at the output of the battery, it conducts when its cathode is at lower potential level than its anode. If for some reason cathode is at higher potential like when the psu outputting higher voltage for a brief moment, then diode stops conducting, and for that moment psu is the main and only dc supply provider.

"Itsu said:..makes no sense to me as the battery is always connected to the system, so why would it "stop powering the system" in the first place?"

Does it have any sense now?

Hi Jeg,  guys,


yes it does make sense now,   thanks.

Itsu

[Void]

That is strange considering the 200x ratio between 5kHz and 1MHz.
I'd look into the scope's sampling frequency relationship.
That is strange, too.
These diodes have a reverse capacitance. Eliminating parallel capacitance would shift up the frequency...unless it was series capacitance.  But where?
What happens when you put a 100pF cap in place of these diodes?

Hi Verpies. I tried putting an air variable capacitor which has a range of about 18 pF to 480 pF
in place of the diodes, and varied the capacitor through its range, and the frequency of the
peak did not shift at all. The frequency of the peak does shift up when the diodes are put in place however.

There is a little bit of ringing on the drain, but that same ringing is there for all frequency settings
of the PWM through its frequency range, but the spectrum analyzer feature on my scope only shows
that prominent peak at certain frequency settings of the PWM as mentioned previously. It is possible that what
I am seeing is just an artifact of my scope however, as it is an economy brand scope. Since what I am seeing is
not related to the ferrite core, I think it is not worth spending more time on it. Thanks for your feedback on this.

If there is in some cases some release of EM energy from a ferrite core at certain PWM drive frequencies, then it may
well be at much higher frequencies such as the high MHz range or even higher. Akula showed a video where he was showing a
peak in the very high MHz range, as best as I could gather from his video anyway, and in another video he mentioned
a frequency in the GHz range. Such frequencies are way beyond the capabilities of the FFT feature on my scope, and
since I don't really trust the FFT feature on my economy scope too much anyway, I will try to use other approaches now.

[T-1000]

Video here:  https://www.youtube.com/watch?v=Zs2r5VI8FH4&feature=youtu.be   

The voltage fluctuations now are raising even more questions:
1) Is your power supply limited to work in 190-240V input voltage range? Seems like that which have a problem. The PS has to work from 100 to 300V input voltage range.
2) As soon as your power supply is switching on, it starts to draw current. That results lowering inductance and the resonant frequency goes up. Is it going outside of the pre-set resonant condition frequency range?
3) Can you attach PLL with current sense on 3T near yoke instead of TL494? This will keep resonant frequency locked in.

Cheers!

[Dog-One]

1) Is your power supply limited to work in 190-240V input voltage range? Seems like that which have a problem. The PS has to work from 100 to 300V input voltage range.

If your power supply is marked "Universal Input", you should be okay.  Attached is the spec sheet for my power supply.  Do notice the inrush current.  It is quite high and can certainly knock the system out of resonance if not limited in some way.

2) As soon as your power supply is switching on, it starts to draw current. That results lowering inductance and the resonant frequency goes up. Is it going outside of resonant condition?
3) Can you attach PLL with current sense on 3T near yoke instead of TL494? This will keep resonant frequency locked in.

This looks like a problem we still need to adequately solve.  I used a PLL connected to the "sync" input (pin 3) of an SG3525 PWM chip.  This does work.  It will give you a resonance self-adjusting push-pull oscillator that will track with the output conditions and still allows you manual adjustment of duty cycle.  The problem I have with this approach is the duty cycle remains fixed during resonance lock.  What this means is the frequency is adjusted AND the pulse width is adjusted.  I'm not 100% sure this is what we want.  I think the pulse width should remain constant or there should be another feedback mechanism to adjust this on-the-fly, like for instance locking it to a specific waveform.  From what I have observed, the pulse width of the PWM may be even more important than the frequency of those pulses.  I'm confident of this being true when driving the Tesla coil and suspect it also to be true when driving the induction coil and grenade.

What it looks like to me is we need a circuit that can manipulate not only frequency, but pulse width and phasing between push and pull sides.  Something that controls on-time/off-time for each side as well as controlling dead-time between the two pulses.  How these need to be controlled to maintain proper resonance is still a mystery to me, but appears critical to proper operation.  I'm currently working on a circuit design that generates a manually adjustable pulse width, then toggles this pulse through a flip-flop sending it to the high side, then low side in a cyclic fashion.  Still working on the dead-time portion to make it adjustable as well.

Controlling this device has three stages that each need their own logic:  Startup, run & variable load compensation.
Maybe this can be simplified at some point.