Kapanadze Cousin - DALLY FREE ENERGY

[itsu]

Look, your scope has 5 channels now!
The 5^th channel shows the current flowing through the Drain connected to the OPPOSITE winding.

Anyway, take a look at the currents in the "gray zone".  The TVS current rises rapidly and it decreases slowly (over ~2μs) while the negative drain current is increasing. (negative increase looks like falling visually)

This is an illustration of unusual current sharing between the TVS diode and the MOSFET body diode.  It is unusual because the forward voltage drop of the body diode is 1.2V but the clamping voltage of the TVS diode is 54V, so the diode with the lower voltage drop should bear the majority of the current, but that is not what happens in the "gray zone".

Could the body diode be that slow?

Do you have a 100V Schottky diode that you can put in parallel with the opposite MOSFET's body diode, just to see how much speedup we can get by helping the body diode?

Bat 46 diode across the other MOSFET drain / source (anode to source, cathode to drain):

[verpies]

But for my understanding, your grey area is not really our problem, the problem is the light blue area where the MOSFET is already on while its gate is off.

In a way because the blue area begins with the gray area...in the opposite primary winding half (and the other MOSFET).

Do i understand correctly that the fact that yellow MOSFET is on (light blue area) is caused by the grey area effect caused by the other MOSFET?

Yes.

The confusion is caused by lack of TL494 output designators.  Lets call them Channel A and Channel B for the future.

[verpies]

First screenshot is zoomed in on the gray area.  I fail to see  the body diode's forward voltage -1.2V there.

Because you should zoom in vertically (even up to the point of clipping) in the area that I have marked in a white ellipse.
The forward voltage drop of the body-diode (V_F) can be different.  The Datasheet states -1.3V max, so it could be e.g. -0.8V

[itsu]

Right,.....    -1.2V it is,   amazing!    (this is with the Bat46 still in on the other MOSFET).

[verpies]

Right,.....    -1.2V it is,   amazing!    (this is with the Bat46 still in on the other MOSFET).

OK so putting it all together:

Ch0: Voltage across the disconnected primary half (Winding B)
Ch1: Voltage between Source & Drain connected to Winding A
Ch2: Gate voltage of the MOSFET driving Winding A
Ch3: Gate voltage of the MOSFET driving Winding B
Ch4: Current through the TVS diode connected to Winding A, when Winding B is disconnected.
Ch5: Current through the TVS diode connected to Winding A, when Winding B is connected.
Ch6: Current through the MOSFET driving Winding B, when that winding is connected.

But for my understanding, your grey area is not really our problem, the problem is the light blue area where the MOSFET is already on while its gate is off.

Let's revisit this issue

Why do you think that the MOSFET is ON?
Let me guess - You probably think that it is ON because the voltage on its Drain is LOW.
...but Drain voltage is not a good indicator whether the MOSFET is ON !
The proper indicator is the current flowing through the Drain of this MOSFET.  Notice that this current is ZERO in the blue area!

Do i understand correctly that the fact that yellow MOSFET is on (light blue area) is caused by the grey area effect caused by the other MOSFET?

That question should be rephrased:
"Do I understand correctly, that the fact that yellow Drain voltage is LOW (light blue area) is caused by the grey area effect caused by the other MOSFET?"

The answer to that rephrased question is: "YES"

I anticipate the next question:
If the MOSFET B is not responsible for driving the voltage on its drain LOW, then what is?"

The answer to the anticipated question is:
"The 1:1 autotransformer effect transforming the positive "flyback" voltage spike on Winding A into a negative voltage spike on Winding B.  That negative voltage drives the Drain of the B MOSFET  LOW"

Note that, at the beginning of the gray zone, there is a lot of magnetic flux stored in the core and the leakage inductance.
When MOSFET A turns OFF, the MOSFET B has already been OFF for a long time.  Thus there is no path for any current to maintain that flux in the core, and the inductance attempts to prevent the change of flux.
The inductance attempts to maintain the flux and the current generating that flux, by finding ANY path for the current to continue to flow.  Thus the inductor increases the voltage on the Drain of MOSFET A until something breaks down and allows the current to continue.  Fortunately there is a TVS diode there to break down at +70V.  If the TVS weren't there, then the MOSFET's A Drain/Source junction would break down...or air would spark.
Anyway, once the TVS diode starts conducting, the current continues flowing through Winding A but it encounters a lot of resistance to this flow because this TVS diode has a 70V voltage drop (so a 1A current would dissipate 70W in it !). Such dissipation represents a huge energy loss and as a consequence the current and magnetic flux in the core decrease rapidly.

The rapidly decreasing flux in the core causes a negative voltage to be induced in Winding B by the 1:1 autotransformer action.
You can see that negative voltage as -15V on Ch0 in the scopeshot below.
Note, that at the beginning of the gray zone, Ch1 jumps up 70V and Ch0 drops down 65V.  That's almost the same voltage change illustrating the 1:1 transformer action (it is not exactly the same because the transformer is imperfect).

On that scopeshot, the -15V is measured on the disconnected winding B.  However when this winding is connected to the MOSFET B then its body diode starts conducting and clamps that -15V  to -1.2V.
This voltage (-1.2V) is so close to 0V that at low vertical magnification it looks like 0V and makes you think that MOSFET B is ON because the voltage at its drain is 0V.  But that is an illusion! 
In fact, MOSFET B is OFF (there is no current flowing through its Drain) and the LOW voltage on its Drain is caused by the autotransformer action explained above.

I anticipate a next question:
"How long is the autotransformer action capable of driving the Drain of MOSFET B LOW ?"

The answer is: "As long as there is magnetic energy stored in the core"

A negative voltage is induced across Winding B as long as the flux in the core is decreasing.  Once the magnetic energy is exhausted and the flux stops decreasing (because it reached zero), the negative voltage stops appearing across Winding B and on the Drain of MOSFET B.

As you decrease the duty cycle, you can see that energy getting exhausted and the voltage at the Drain jumping back up before MOSFET B really turns ON.  This is because shorter pulses store less energy in the core, and less energy lasts for a shorter time.