UCC28950: SLUSA16D RC SNUBBERS ARE NOT USED?

Hi Samrat

The PSFB is a zero voltage switched topology and doesn't normally need snubbers on the primary. When the 'bottom' switch turns off, the voltage swings positive where it is clamped as the body diode of the 'top' MOSFET turns on. When the top switch turns off then the voltage is clamped by the body diode of the bottom MOSFET - snubbers are not needed because there is no unclamped inductance available to generate the large spikes. This is in contrast to the Flyback topology for example where the leakage inductance energy has to be absorbed by a snubber in order to prevent it generating high voltage spikes.

Having said that - the assumption here is that there is no unclamped inductive energy in the H bridge. This is true if the PCB layout is good but any stray inductance in the source or drain circuits of the MOSFETs will generate spikes - the solution is normally to improve the PCB layout but a snubber would also work.

Hope this makes sense.

Regards
Colin

hello colin,

Suppose I am using deadtime more than what is required for ZVS (not expecting ZVS) still

'''When the 'bottom' switch turns off, the voltage swings positive where it is clamped as the body diode of the 'top' MOSFET turns on. When the top switch turns off then the voltage is clamped by the body diode of the bottom MOSFET'''

this happens right?
so can i say this property of FULL BRIDGE ? or am I missing something?

Thanks
samrat

Hello Samrat

You are correct - providing of course that there is enough energy available to drive the ZVS transition to completion. Note too that the energy available to drive the PA transition (QA,QB switches in UCC28950 DS) is less than that available to drive the transition in the AP leg (QC, QD switches in UCC28950 DS. You might want to look at the Infineon design note at www.mouser.com/.../2-12.pdf which talks a lot about the energy requirements to achieve ZVS.

The ZVS action is a property of the PSFB because the topology - or more accurately the switching pattern - preserves energy that can be used to drive a ZVS transition. The alternative PWM Full Bridge doesn't provide ZVS.

Regards
Colin

hello colin

but suppose fixed deadtime is longer than what is required for ZVS and shim inductance is also not there (enough energy is not there to charge two COSS output capacitances ) in this still this control strategy is better than normal PWM or all the advantages are lost? then how turn off snubbing is achieved.

and is there kind of partial ZVS also possible? if energy is not enough?

thanks
samrat

Hello Samrat

First - yes, partial ZVS is possible. In fact it is used in quasi resonant Flybacks to reduce switching losses - in this case it is called valley switching. For example if the switched node has 50% completed its transition when the MOSFET is turned on. In this case you will get switching losses but they won't be as great as if you had hard switching from rail to rail. Remember that switching losses are proportional to 1/2 C V^2 so if V is halved the losses reduce to 25% of the hard switched value.
Timing is important though - if there is not enough energy to charge the Coss fully then Coss will resonate with stray inductance and you will see a decaying sinusoid waveform at the switched node. If you catch this waveform at a minimum (valley switching) then you get lower losses than if you catch it at a maximum.
I think this kind of partial ZVS would be difficult to control - simply because of differences in timing between different controllers and power stages.

Turn-off snubbing isn't normally needed on a PSFB but you can look at examples from hard switched bridges to see how it might be done.

Regards
Colin