little update: I got some new gate drivers and hooked up the circuit as discussed in this thread.

The result was that my converter fried again, but my gate-drive opto made it out alive - thanks to the gate resistance which i forgot last time.

The circuit works fine when the fet is unloaded. By that i mean that i can see the gate drive waveform behaving nicely without my bulb connected to the drain.
Using my scope i can see the rise is much sharper with the gate driver and the time before passing the miller plateau is greatly reduced compared to driving the gate with a v-divider.
Seemed good so far.

When i connect the bulb to the drain however, then increase the duty - the lm2596 chip on my board smokes. But as i said, unlike before - the opto made it out alive.
I had the input of my converter connected to a cap that is charged to the peak value of my 25V rms supply which is causing the issue i believe.
Is it reasonable to believe that the effect of the load switching is causing a voltage spike which is frying my converter?
The switching frequency is 20kHz, so am i switching at a speed that doesn't allow the loading effect on the transformer to reach equilibrium?

 

Is it possible to try with a smaller wattage lamp, and then check the amplitude of the spike at LM2596 input? If it is the spike, then the reduced wattage bulb, will have lower current, thereby lowering the spike voltage, which could be less forgiving on the chip.

 

i dont have the option of using a lower wattage bulb, i am using a 150W now and actually plan to go higher.
At the moment i am using a 12V zener diode to supply my gate drive opto.
It works well, but my 3W resistor in series gets really hot, 100 celcius +, but that will be ok when i heatsink it.

 

I was thinking of lower wattage bulb only to confirm if it was the spike at the input of LM2596 that was blowing it, as it was blowing immediately when the duty is increased.

 

I was thinking of lower wattage bulb only to confirm if it was the spike at the input of LM2596 that was blowing it, as it was blowing immediately when the duty is increased.

I actually used my friends fluke multimeter, it shows that the peak voltage of the transformer is around 38-39V.
So If switching my load caused even a small spike, then this 40V level was exceeded = SMOKE.

You were correct RamaD.