How to wind a push-pull transformer?

Started by Piotr Wyderski August 4, 2018
<> wrote in message
>A 50% duty cycle is incompatible with break-before-make switching.
>It takes a finite time to switch a MOSFET on or off - you've got to move a >finite amount of charge into or out of the gate electrode - so a 50% duty >cycle implkes that one is turning off (but not turned fully off) at the >same time as the other one is turning on (but not fully turned on). >
Yeah, for PP, you have to deal with the transformer. Small deadtimes (including negative) can be okay for half bridge, as long as you're not driving a low-DCR load (like a transformer without a coupling cap). The classic solution being the ATX power supply, where a half bridge drives the transformer primary, returned via 2.2uF film cap to the FWD supply's middle tap. (The cap could return to any end of the supply, but the middle tap is convenient to reduce the startup transient. Alternately, a "half bridge" of caps can be used -- a capacitor divider -- which is the preferred method for FWB supplies.) Note that, within a cycle (i.e., ignoring "flux walking" bias over many cycles), each switching transient involves the inductance between switches. In the PP case, that's the end-to-end leakage. In the half bridge, it's stray wiring inductance, from nearest bypass cap, through the two switches. Zero dead time, or slightly interleaved, switching is possible when that loop inductance is intentionally controlled, setting dI/dt. This is preferable for synchronous switching, and for bidirectional converters (power converters, class D amps) where consistent EMI and higher efficiency is needed. (Not much higher efficiency, mind -- the high frequency reactive current drawn by the overlap has to be dealt with appropriately after all. The main thing to be gained is body diode recovery, which is a monster with higher voltage MOSFETs.) Also, a current-sourced inverter MUST operate interleaved. In that case, "dead time" is dead in the voltage sense, meaning, both devices ON during the dead time. This isn't used much in power applications, but is relevant to some configurations ("Royer oscillator") and RF amps (e.g., class E, PP). Tim -- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website: