Nick Hi had a look at your Scop shot of your yoke photo, If that's your device I would say your transistors are in an 'not on or off but in a resistive mode if I were you I would go for a different winding to try and minimise the effect or get a better 'core' you want the ringing in the transistor off time or dead time, I sincerely hope this helps you.

PS I had this problem my self.

Allen

Nick,

Although your dead time control on the PWM is OK, the 'effective dead time' may not be and this can be checked with drain to source current waveforms as suggested by Itsu.

   Itsu: just how should I test for the current wave form?

    Hoppy: No, I have not connected the feed back circuit up, yet.
   There is no point in connecting the feed back circuit until I resolve the snubber heating issue. The induction circuit need to be able to work well by itself, first.

  Itsu: Is it possible to show in a short video of your induction circuit lighting 400 watts worth of bulbs, for a few minutes run time? As well as showing the lumin output at your bulbs, and the temperature of the snubber resistors?

    Hoppy: No, I have not connected the feed back circuit up, yet.
   There is no point in connecting the feed back circuit until I resolve the snubber heating issue. The induction circuit need to be able to work well by itself, first.

OK. I assumed from your previous comment:" Adding another load by connecting up the feed back path is not going to lower the heating, but raises it instead." that you had done this.

   Itsu: just how should I test for the current wave form?

    Hoppy: No, I have not connected the feed back circuit up, yet.
   There is no point in connecting the feed back circuit until I resolve the snubber heating issue. The induction circuit need to be able to work well by itself, first.

  Itsu: Is it possible to show in a short video of your induction circuit lighting 400 watts worth of bulbs, for a few minutes run time? As well as showing the lumin output at your bulbs, and the temperature of the snubber resistors?

Nick,

as Hoppy already mentioned, you can test the current by using a current probe, or a CSR (current sensing resistor).
I think the best way is to put a non inductive 1 Ohm resistor in the MOSFETs source line, so you can use your normal probe to measure the voltage (thus current) across this csr.

See the below video of my setup running for some time using 400W of bulbs (220V)
It also shows the current probe being used which was set at 5A/Div.  (second part of the video!!)

Nothing is really getting hot, but there is not much luminense on the bulbs either.
All temperature readings are in degrees celcius.

Video here:  https://www.youtube.com/watch?v=hSfNrwqkPhM     (don't use IE to display).

Screenshot shows the:
MOSFET-1 drain voltage (yellow),
drain current (green @ 5A/Div.),
gate signal (purple) and the
MOSFET 2 drain voltage (blue).


EDIT: i realize that the green current drain trace was taken from a primary coil lead BEFORE the Oleg snubber diode, so it does NOT really represent ONLY the drain current.


Itsu

   Itsu:
   Ok, thanks for the video and advice. 
   Without the 28t coil and 168t coil and rectifier hooked up, it's hard to properly judge or compare the results.
   Your lumin levels would probably much higher on those bulbs if the 28t was also connected up. And possibly the heat on the snubbers, as well. In any case,  the heat on the snubbers (or lack of), was what we were really trying to show and understand.
So, it does look like the cleaner the square wave signal becomes, the less output it will provide.
   I do very much appreciated you helping me out in this way.

   Hoppy: 
   First I'll install the new carbon resistors that you recommended. Which I have now received. They only come to 1/2 watt size, so I'll have to use about 4 of them in series for the Oleg snubber, to get the 2 watts of resistance. And we'll see if the RC snubbers 1/2w 47ohm carbon resistors hold up, or not. I may have to double them up, as well.
 
  I have a lot more voltage and current going through my device, as compared to Itsu's test and video. My normal peaks with snubbers are over 200v at resonance. And at avalanche can get to over 350v. Which can partially light 600w worth of bulbs, or higher wattage, and this without going into the avalanche effect. So, that is probably why my snubber resistors get so much hotter.
But,  something may not be quite right on my set up, still. In any case, we'll see what happens when the new carbon resistors are installed.

  I now also have 10 more brand new Mosfets to play with. As well 4 new 4420 drivers chips of the right (through hole) size,
  Once I install the new resistors, I'll make another short video.

   Hoppy: Ok, then I'll install them in series. Error corrected now.
   Thanks for reminding me.

   AG: You mentioned:
"Nick
No offence Nick 'mate' but why doesn't your waveforms look like the one above?
                                                                                                              end quote.
   No offence taken, mate.
   All I can say is that that is like comparing apples to pears.
   Why don't you build a proper Ruslan/Akula/Stalker type of replication, then we can compare notes.
As as I've been going with Itsu.

Well, what's the difference all the devices work under the same principle unless you're referring to the yoke?

Well if you built a device with a yoke that yoke has to be a Russian yoke of military speck what I'm asking you is, is your and Itsu's device a Russian military spec device because I couldn't get one?

if it is where did you get it from and you need to ask your self why doesn't it work in your device correctly?

Allen