Note that the above calculation is true, using the energy left over after the secondary has "caught" the flyback -- in other words, the energy stored in the primary referred leakage inductance.
Which is charged to the same current as the magnetizing inductance, at switch turn-off, but the inductance is a small fraction, say a few uH (check the transformer datasheet, or measure it if you don't know -- this is the primary inductance with all other windings short-circuited). So the RCD components don't need to be very powerful, at least assuming a reasonable transformer design.
The parallel R and C I think aren't needed, or their presence varies between drawings/applications. The zener/TVS does the job, as long as it can handle the power, and does so at a stable voltage drop. Whereas the RC case, voltage varies with load, affecting efficiency -- the voltage is never less than the reflected output voltage, so at light enough load, switch current must be at least enough to overcome that loss. The series resistor I haven't seen very often, but its value is small enough it shouldn't do much to conduction loss (manifest as excess switch peak voltage), meanwhile it significantly reduces EMI by damping the diode capacitance. You may also see a ferrite bead (on the diode, or as a separate chip or leaded component) to the same end (adds ESR at ~VHF). (Both diodes are necessary (the rectifier and TVS), because a unidirectional TVS alone of course won't do, and a bidirectional may conduct during the on-phase, at high enough input voltage; but the rectifier also has much less capacitance than the TVS, which would otherwise affect switching.)
There's also the dV/dt (rate) snubber, arranged slightly differently and using a much smaller RC time constant (and no TVS); I don't think this is used much, or very important, for a small regulator like this, but it's an option where the load current is generally high (so that the rate is nominal; it's proportional to peak switch current, of course), leakage inductance is relatively low (since a relatively large RC is needed to absorb the peak overshoot of larger LL), and the effect is to reduce turn-off switching losses (which can increase overall efficiency a little bit). Its value can also be dimensioned to dampen free ring-down (in DCM converters), further helping EMI.
Tim
Home
Help
Search
Login
Register
