[zilog]

What current and power loss must the antiparallell diodes that prevent body diode saturation in a full bridge configuration handle? I am thinking about designing a subwoofer driver stage that operates from 45V and is able of driving a 2-4 ohm load, thus pulling 20A top. How to do these calculations, how do I model a typical worst case speaker (I guess one benefits from knowing inductance and resistance of the speaker) in order to calculate the amount of energy delivered to positive rail during each switch-off. Is there any possibility of handling these requirements using surface mount diodes, or is heatsink mounted TO-220s my only option?

P.S. Entschuldigen Sie bitte meine Sprachewahl, Ich habe jetzt Urlaub und deshalb benutze Ich keine Deutsch

[alfsch]

back-diode gets full load current, but only for short time.
parallell diode is almost useless, internal diode switches on anyway...
so better, to have good timing and mosfet with fast back-diode inside

[zilog]

You mean to use mosfets with diode besides the body diode built in? I have a large pile of IRF540's, are these completely useless? I used them in my half-bridge prototype anitparallelles with fast schottky diodes with good results and with very clean switching waveforms. Unfortunately a blown plastic capacitor in the output filter killed it, with complete set of op-amps etc, so I cant do any more measurements on it.

[alfsch]

no, just the body-diode only; there are newer fet like FDD3682 , 32A, 100V, 30mohm, diode trr 55ns !

with irf540 , trr 500ns ... adding a schottky could help, but did you try without extra diode? switching speed?
can you show the circuit?