advanced Dremeling

Looks like a forward converter. You don't usually use beads on the output of the diodes, you need a regular inductor on the output and a freewheeling diode.

Cheers

Klaus

I wouldn't call this a forward converter. It's an unregulated, square wave, dc/dc converter.

With a bridge rectifier driving a capacitor and no inductor, there's nothing to need a freewheeling diode. The ferrite bead only softens up some fast dI/dT edges and reduces ripple a bit but doesn't store much energy. It's tiny compared to a forward-converter inductor.

Yeah, that's right, it's a special case with 50% duty cycle, the conduction time of the secondary bridge is close to 100%, so not much to filter out.

Just beware, that a bead is also pretty inductive at lower frequencies, so if you do not need the resistiveness (if that's a word) of the bead at +100MHz, a regular inductor could be a better choice.

That 5 amp bead is about 2.9 uH at low frequencies, enough to soften the driver side current spikes some. The HF loss damps ringing during the half-bridge dead time. The Coilcraft planar-winding transformer has very little leakage inductance so this circuit can have giant current surges. This is the board that deprogrammed its own FPGA at power startup.

The PCB layout had too much current loop area too. It's hard to explain these sorts of things to our layout guy, and we should have thought about it more. Rev B should be better! We're brutally hacking a couple of rev A boards for the customer demo.

I have, coincidentally, essentially the same thing going on on another project, the low power one with the charge pump supplies off to the side of the main dc/dc conveter. Switching between those can be confusing; I hate to context switch.

I used to build CD ignitions for motorcycles, and never really understood how they work. The coil primary was in series with the storage cap, which softened up the inverter charging for free. That was pre-Spice as my excuse. I bet I can remember the circuit from long ago.