Hello

Recently i started studying AT,ATX circuits, (example circuits attached) and made some smps ATX like smps, because of what, now I regret it (mosfets much better ) ).
Power transistors like MJE1300(7/9) overheat and it's obvious that turns count and maybe core without gap is the reason.

Playing with positive feedback winding turns count (increasing), led to self-oscillation and ensuing power transistor damage , but in general this winding is a protection winding, the more current through power transistors, the more base current and less Vce drop.

How correctly design this non-standart push-pull flyback converter: primary, secondary, feedback windings, core selection ?

 

So... Any transformer, including the ferrite core transformers, underlay the formula (turns/Volt)=50/Sc, where Sc is cm2 BUT there must be applied the frequency correction factor, because this formula is right for 50 Hz. Thus, if You have 50 kHz, must turn 1000x less turns. And must be made the correction factor of ferrite B(max) because formula is designed for 0.9 Tesla whilst most of ferrites permit 0.25...0.35 Teslas. Thus, proportionally to datasheet You must wind little bit more. For example at 0.3 T must wind 3x more. Now correction to flyback trafo. Flyback must be calculated the same way as normal transformer, but it is much more sensitive about Bmax. Thus, I cannot to see any other good method as gradual iterations winding slight more or less whilst measuring the oscillo of current in the bobbin. Early warning - never use the inductivity method to analyze the trafo, because the L is sharply variating as value dependant on actual value of B. You may find the zero power inductance or short-circuit load L (what is by definition zero or more exact the r of wires).

 

As soon the ait gap is involved then formula is deploying toward (ampere*turns)=Bcore*[Lmag by ferrite/mju(rel)/mju(zero)+L(by air)/mju(zero)] where mju(zero) is pi()*4E-7