Snubberinos and SMPS secondary side.

Is there a choke after the diodes or they are connected directly to the reservoir capacitor?

You are dealing with the spike in the reverse direction of the diodes, is that right?

Those two options could work.

1. Punk solution: slow down the turn off of the FETs on the primary side.
2. Use the full bridge rectifier at the secondary.

1. Non-dissipative L-C-D snubbers, or additional winding on the transformer.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

Yes there is a choke after the diodes with the traditional LC output filter.

I'll look at slowing down the fets. I hate doing that but it may be the lesser of two evils.

Full bridge no possible. :(

Tell me about implementing LCD snubbers in this case.

thanks Vlad.

I wonder if one could build a Cockcroft?Walton voltage multiplier sort of thing to capture the ringing and recirculate the energy into the DC output node. It actually wouldn't take a lot of parts... some Central multi-diodes and some capacitor packs.

Or maybe highpass it, transformer couple, and return the energy to the primary side. The recycler could even be resonant at the ringing frequency.

John

I was thinking the same thing of stearing that energy back to the input or the output. The input and output are on the same ground (non isolated) so thats a step in the right direction.

The ringing is normally associated with the arm that is turning off - in this case freewheeling current being interrupted in the rectifier that is being temporarily reverse-biased.

If the snubber is dealing directly with this single loop, it is more effective, so two individual snubber networks will work better than just one. As the voltage swing on the snubbed branch is halved, the capacitive energy loss in a single snubber is quartered - net energy loss is halved for the new pair being used.

Lossy 'spike-killer' type beads might be a prescription if the oscillations were ten times higher in frequency, but at 3MHz I see no benefit, except perhaps in reducing the reverse di/dt in the extinguishing rectifier. At higher voltages, the bead would be in series with the rectifier and a small cap placed across the rectifier itself. As the beads saturate with each switching cycle, power loss may be currie-temperature limited, unless adequate heat-sinking is provided.

RL

The ringing could very well be caused by the kick back of the L into the stray inductance of the transformer. Try adding the catch diode to the ground before the LC, just like in the buck topology.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant