Offline 70W Phase sift Full Bridge

[cupoftea]

Hi,
We are doing a 70W, 100kHz Phase shift Full Bridge for 180-265VAC input.
Current doubler output.
24V, 3A output
(Parallel 3 output diodes)x2 as synch rects not worth it for 3A.

So why isn't anybody else doing this?

Pri FETs will be 4 x DPAK (+ turn off snubber for Bottom FETs)
Transformer nice easy wind, only the pri and sec coils and no split sec.

So this is easy, and we wont even need any heatsinks, even though it will be in
totally enclosed plastic with no fan.

So why isn't anybody else doing this?, I mean, no heatsinks, easy transformer wind.

Design is no more difficult than a "plain Full Bridge"

We will only get ZVS turn on at above 70% of max power but that's fine.
We don't want an extra "leakage inductor" as it just means diodes to the rail also needed
and also more dissipation in the output snubbers.

 

[Easy peasy]

possibly because at 70W a valley switched flyback is less parts and similar efficiency ?

 

[cupoftea]

Thanks, though that would need heatsinking, and need at least a three coil transformer (interleave wound). With the PSFB, we can do it with no heatsinking,
and with all SMD parts, apart from maybe the common mode choke and HV Bus cap and transformer.
So anyway, the lack of heatsinking and the assembly effort of that makes the said PSFB well worthwhile we believe.

 

[Easy peasy]

It will be interesting to see if what you say does indeed turn out to be the case . . .

please let us all know the outcome.

--- Updated Jan 9, 2025 ---

This part here would dissipate < 50mW in a 70W flyback, valley switched design, for 250VDC ave in

https://octopart.com/datasheet/ipb65r065c7atma2-infineon-93259729

3.60 USD

--- Updated Jan 9, 2025 ---

Or two of these in parallel ( each with 4E7 gate R ) 125mW each and only 45c USD each

https://octopart.com/datasheet/ipd70r600ceauma1-infineon-70523113

ground ref GD, lots of chips to choose from - no bootstrap needed, diodes can still be // for the output and heat-sunk to the pcb quite effectively - 3A x 0.8V say = 2.4 watts ave, 3 diodes in // = 0.8 watts each so 3 x SMC package diodes with plenty of copper should suffice - or leaded.

EP

 

[cupoftea]

Thanks, another point is that we may actually elect for the PSFB with external "leakage" inductor in the primary side...and then have diodes to the rail between this external "leakage " L and the transformer. We would then sandwich the secondary between two series primary halves. This would reduce the ringing on the output diodes. This will be useful because a high leakage term will ring like mad with the Cds of the synchronous FETs that we will end up actually using for this , and a future 24V, 10A product which has just now been given the go-ahead. So since the future product uses synch rects we might as well use them on the 70W product aswell.

It surprises me that none of the PSFB app notes actually mention that if you are also using synch rects with PSFB, then you really do need to interleave wind the PSFB transformer, and then have the external leakage inductor with diodes to the rail.

PSFB App note.

https://www.mouser.com/pdfdocs/2-12.pdf?srsltid=AfmBOooi5bGWT4KxZrKTgHrnQXNUREEDerLbzuvtYAFRPy54mUXM9lF1

Yes another app note below correctly suggests the use of a current doubler rectifier for a PSFB_with_synch_rects.....but says nothing of the need to use interleave winding of the PSFB transformer.....Current doubler output is THE way to go with PSFB when synch rects are used because the transformer has a single secondary coil (no split sec), and so this can easily be "sandwiched" between 2 primary halves to give v low leakage inductance in the PSFB transformer.

Why does not a single app note on earth today actually report this highly important point?

PSFB app note

https://www.ti.com/lit/an/slua287/slua287.pdf?ts=1737554383701#:~:text=Phase%20Shifted%20Full%20Bridge%20Switching,%2C%20QB%2C%20QC%20and%20QD.

 

[Magnethicc]

Thanks, though that would need heatsinking, and need at least a three coil transformer (interleave wound). With the PSFB, we can do it with no heatsinking,
and with all SMD parts, apart from maybe the common mode choke and HV Bus cap and transformer.
So anyway, the lack of heatsinking and the assembly effort of that makes the said PSFB well worthwhile we believe.

So you would use additional 3 switches and additional 2 output chokes to... save on heatsinking?

 

[cupoftea]

Thanks yes, because of the chance that an assembler may not fit and mount the heatsink properly. Its such a manual task.