Full Bridge is Dead....Phase Shift Full Bridge replaces it?

[cupoftea]

Hi,
Is there any need for anybody to ever do a Full Bridge SMPS any more?
Surely you just do a Phase Shift Full Bridge instead?

Take say a 240W, 24V, 10A Full bridge from a 400VDC PFC output.....just do it with a cheap Phase shift full bridge controller such as UCC28951.
It even has the synch rect drivers on it aswell.
There are no offTheShelf Full Bridge control chips with sync rect drivers on them.

So just use the Phase shift full bridge, and dont bother with any extra "leakage" inductor....just wind a layer of secondary over a layer of primary (no interleaving) and youre done......who says a Phase shift full bridge needs to be soft-switching in leading and lagging lags at full power?..........The PSFB done thus will be at least as good as the Full Bridge in this case of 240W output.

I mean, put the extra inductor in there if you will , but no need at 240W. Save yourself a part...its only 240W output.

The two attached sims show PSFB "done like a full bridge" is the best way forward.

The constant duty cycle switching also means none of the "duty cycle change" problems in any gate drive transformer.

UCC28951

UCC28951-Q1 data sheet, product information and support | TI.com

TI’s UCC28951-Q1 is a Automotive green phase-shifted full-bridge controller with SR control for wide input voltage range. Find parameters, ordering and quality information

www.ti.com

 

[cupoftea]

However, the efficiency of the PSFB comes at one heck of a cost.
There are several failure modes which are far bigger in the PSFB than in the standard Full Bridge.
These failure modes are as follows...
(the failures are most likely in light load, in short cct output, overload, and startup/re-startup.)

1...PSFB is prone to fail due to reverse recovery of the FET's internal diode.
2....PSFB is prone to fail from spurious turn on of the "other" FET in the leg. (due to VGS getting induced up)
-to stop the spurious turn on, the PSFB FET needs low Rg
-also to stop the spurious turn on, the PSFB FET needs low Qgd/Qgs.

The plain Full Bridge does not suffer this spurious turn on nearly as badly as the PSFB..nor the reverse recovery issue.

If you ever do a PSFB, you absolutely must update the PCB number to indicate exactly what FET part number was used on it.
This is needed because then you can see the number of failures you get with that particular FET.
Infineon and ST and Vishay do special FETs for PSFB which try and mitigate the failure issues.
However, there can be no true cast iron guarantees.
There is no way of calculating that a certain exact trr value will give no failures.

The PSFB needs somewhat to be used in applications where it can be "wrapped in cotton wool"...and if your application involves the customer
having access to the power supply output, where they may short the output, then do not choose a PSFB unless you want a lot of failure returns.
Choose a standard Full Bridge and just use a low enough switching frequency so you can get away with a damped drain node transition.

(These things also apply to the LLC converter)

So yes....some applications for high power supplies use a PSFB so that they can get it small....and they tolerate the higher failure rate that the PSFB will give them.

But if your application involves plentiful overloads, startups/re-startups/short circuit outputs...then be wary of choosing PSFB.

We used to repair 3kW Electro plating power supplies....no way did they use PSFB or LLC. They had 6 diode , 3 phase inputs.
No way did they choose PSFB or LLC because there was a high chance of customer applied short circuit.
They just used simple Half bridge converter. With no output inductor and just used the high leakage inductor to limit output current.

Take a look at your server farms......yes they use PSFB and LLC as the amount of them means efficiency very important....but they are in racks with redundancy so the failed units can constantly be taken out and replaced without the server being down-powered....this again is a mark of the PSFB and LLC high failure rate. But try doing it with plain full bridge and your 'leccy useage goes through the roof.......plus the size of the server farm when you have thousands upon thousands of PSUs you do need to make them small and just tolerate the failure rate of the PSFB and the LLC. Hot swap them out.

Same with these 800W single phase 240VAC oil rig PSUs that we repaired (fans failed after 10 years).......they used open loop half bridge.....no output inductor but just used the leakage...just switched at 48% all the time.....output wandered and so Buck used to regulate it to 24Vout........all that "mess" just to keep away from the failure prone PSFB and LLC. (they were switching off the 400V PFC output)

I once worked at a teleco place and they needed a 800W PSU......they elected for PSFB but should have just used plain Full Bridge.......there was no great need to make the PSU small as they were in a place where there was plenty of room.....The usual thing where an Apps engineer stands the companys engineer on their pride and says "you arent scared of doing a PSFB are you"?...then the muggins goes and does PSFB.


FET failure modes in the PSFB

https://www.infineon.com/dgdl/s30p5.pdf?fileId=5546d462533600a401535748559c3fcd

Just compare the two attached sims to see the malaise that the PSFB gives when compared to the plain old full bridge......
During overload and Short cct output, the PSFB is seen to suffer both reverse recovery and also spurious fet turn on far more than the full bridge.