300W Pushpull with 15W dissipation in each primary snubber

[cupoftea]

Hi,
We have been given a 300W synchronous Pushpull to bring up by a customer.
24VIN, 32Vout, 225kHz, isolated.
Transformer is 1:2.
Its planar and the manufacturer tells us the leakage L is 1.2uH “measured at primary with sec shorted”.
The attached planar spec, is from the Planar Company that designed this planar transformer for our customer.

The thing is, the peak primary current is 21A. …and with 600nH of leakage L in that primary “half”, that means an energy of 0.5xLxI^2, and a power dissipation from that leakage inductance (in the primary TVS) of some 15W. ( from 0.5 x L x I^2 x f). The simulation confirms this level of dissipation in the snubber (TVS).

That’s way too much as you can tell.

Would you agree, the Pushpull cannot be used for these levels of leakage inductance? There’s nowhere for the leakage inductance current to flow, when a primary FET turns off. I wonder how on earth the pushpull even exists above very low powers. I mean with planar you can (in theory) get very low levels of leakage inductance….but at the cost of large interwinding capacitance. -And we don’t particularly want such large stray C in a hard-switched converter. I believe we will request a re-do as a Full Bridge. Would you concur?

LTspice and jpeg attached

 

[Easy peasy]

you could build a diode-cap catcher on each pri side mosfet and then have another smaller push-pull or half bridge send this energy to the sec side - you can use the same gate drive.

 

[dick_freebird]

Perhaps there is a clever push-pull-specific "snubbage" / clampage
that (say) returns the flyback spikes to the center tap?

I gather from light reading that push-pull transformer design gets
some specific attention. Is there a reason why you get to stick with
a crappy one?

 

[Easy peasy]

They're all crappy until you fix em, even with a turn off catcher and send to the o/p, you still have to deal with the sec diode rectifier turn off - once again a cap catch ckt can be used - but now an active ckt works better ...

 

[BradtheRad]

I can't say whether an active or passive circuit is best, but in case this helps...

"Flyback Lossless Passive Snubber"

www.researchgate.net/publication/308814793_Flyback_lossless_passive_snubber

"A novel regenerative snubber circuit for flyback topology converters"

www.researchgate.net/publication/342567538_A_novel_regenerative_snubber_circuit_for_flyback_topology_converters

 

[cupoftea]

Thanks for these replies.
Yes, the LTC3723-1 controller was used, and in the back end of its datasheet, there are examples of 100W+ pushpulls. Why they would tout such a lousy converter is beyond me.
I believe the pushpull should be condemned. Any datasheet touting it should stamp the page with "warning", like they do with "not for new designs".

Planar:
The top post shows the planar transfomer spec as written by the Planar Designer company themselves....so they knew it was for push-pull useage....and yet still served up a Planar transformer with 1.2uH of leakage between pri and sec.....(there own spec shows it).....they are one of the well known Planar designers....so how they didnt realise this is beyond me......maybe they just thought the myriad of mitigation circuits needed for this, would be used here.

Mitigation:
In my view the correct mitigation of the hard-switched push-pull, is "choose another topology instead". If you hear the word "push-pull"....Run!

Push-pull term:
Bizarrely, almost all Full Bridge, and Half Bridge drive controllers are labelled "push-pull controllers"...you literally have to use "push pull" in your search term, even though you wouldnt want to touch the pushpull with a barge pole.

Inter-primary coupling:
The pushpull , as kindly suggested elsewhere, needs really supertight coupling between each primary half, so that the magnetising current can be kept flowing after a primary fet turns off......the magnetising current then flows through the diode of the opposite primary fet....and great coupling (between primary halves) is needed so that the reset of the magnetising current can proceed unhindered in this way.

LTC3723-1

https://www.analog.com/media/en/technical-documentation/data-sheets/372312f.pdf

 

[Easy peasy]

You can also add diodes to the layout on the pri side to reduce the loop area - to reduce turn off volt spikes a little.