It is very difficult to keep the "Geofusion effect", as well. As most any change in wiring methods can cause it to go away.
Most likely the snubbers are not doing what they need to do, but, if you keep blowing fets, your choke may not be doing it's job.

   My fets are not even getting warm, but when you reach a higher and stronger frequency point of resonance, they can get too hot, also.  The trick is finding the right frequency point, that will allow for the effect, but not overheating anything.  Not easy to do. Even my grenade former tube has bent and sagged in the middle from the weigh of the winding, when the coils get hot. It looks like a hot dog, now.

  apecore changed his fet drivers to be closer to the fets, (right on the fets) like Itsu had done, and lost the "effect". 
He is now using Geos last wiring diagram, to see if some different wiring methods can bring back this effect.
He now has managed some 200+ volts at the bulbs, up from 60v,  so far. But, may still be looking for that illusive effect.
  Verpies:  apecore is the guy that has his fet drivers on the fets, after Itsu's suggestion, not me.  I would not want to risk losing any "effect", at all. 

   I've resized the previous diagram, and have re-posted it here, again.

Quote from: NickZ on November 17, 2016, 08:06:41 AM
   It is very difficult to keep the "Geofusion effect", as well. As most any change in wiring methods can cause it to go away.
Most likely the snubbers are not doing what they need to do, but, if you keep blowing fets, your choke may not be doing it's job.

Nick,

It maybe that your Mazilli is not so prone from interference as Geo's push-pull. Try moving your Mazilli and yoke around to see if the 'effect' is easier to find and maintain.

Quote from: NickZ on November 17, 2016, 08:06:41 AM
It is very difficult to keep the "Geofusion effect", as well.

It is easy once you realize what exactly the effect is
...and design for it.

I learned a lot from you guys, maybe you'll learn from my mistakes
When I saw the effect on my setup I was glad. I saved resonance frequencies, duty cycle, amperage and voltage ..
Duty cycle was 25%. If I decreases duty- cycle the effect disappear and the frequency and the signal was doubled and reverse, upside down.
If i increase duty-cycle , effect disappear to.
I started adding bulbs for load to see how much the load voltage decreases. When it were 600w load, everything stopped. Mosfets were cold but the resistor in snubber was hot and without colors. I realized that higher power means higher spikes and that destroyed my mosfets.
Now I think lossless clamp decreases spikes but might destroy the '' effect ''.
I do not know what solution is best, lossless clamp or snubbers?

Quote from: conico on November 17, 2016, 11:01:59 AM

I do not know what solution is best.

The need for a solution assumes that there is a problem to address. Is the 'effect' a problem that requires a solution, or is it a requirement needed to reach the objective - a self-runner. "Answers on a postcard".