Full Bridge Inverter Snubber Design

I want to discuss snubber design/topology for a phase shifted full bridge adapted for an inverter (DC-1khz) application with about 30A output at ~750w. This is the basic topology with my current simple RC snubber shown:



One key detail is that the back to back switches effectively remove the body diodes from the picture so there is no natural clamping on the secondary. Second, the fact that the output can be AC eliminates some common snubbers for the full bridge topology (Like diode->cap that drains to the output).

The design right now is roughly as follows
400V primary -> 26:2 transformer -> 30V secondary -> 60V secondary blocking voltage + voltage ringing -> ~130V -> 150V mosfets.

This, overall, is workable according to simulation with about 10W of snubber losses but obviously doesn't have much margin. So cutting down on secondary voltage overshoot would have significant benefits in terms of design margin, and/or allow another turn in the secondary for additional voltage capability or allow lower voltage fets with the obvious benefits that brings.


So does anyone have general comments or suggestions for snubber design/topology in this application (or other comments on the desire to limit voltage at the secondary fets). PSFB has plenty of literature but not with a bipolar secondary. I can invent things, but I'd rather not. Note that the schematic as drawn also roughly represents planned layout.

Hi,

Given your experience, not sure/doubt this manual will bring anything new to your table, anyway 4.4.1 - 4.4.4, even if about thyristor and diode snubbers, and 5.9 - 5.? (I'm only at page 260...) about IGBT/MOSFET snubbers maybe could be of use/lead to an idea for what it is you're doing?

As you can guess, I know nothing about this area in practical terms, and very little theory either - I thought SCRs were happy dancing diodes, kicking a leg out, until a few days ago..., so will keep my mouth shut other than uploading the manual.

Consider two DCR snubbers, one in each direction.



My schematic resembles the action of an H-bridge sending AC square waves through a transformer.

Ok I've seen this in DC applications where the 30k 'drains' to the output which has the benefit of recovering some energy. Let me make a few other possibly dumb comments.

Using this in an AC application where I don't see any easy tricks for recovering the clamped charge its a completely lossy snubber (right?) so why wouldn't I use a zener clamp. And in that case I'm basically just moving losses that my secondary mosfets would theoretically absorb in avalanche mode to another device, and/or possibly losing more by clamping at an earlier voltage. From a big picture point of view moving losses around may still be useful, but I feel like I may be missing some other options/variables here.

Hi,

Missing the point completely... Would soft switching (ZVS or ZCS) add anything to what you're working on improving in the design, or does that have nothing to do with the issue at all?