Can the UCC28950 be scaled to a 6kW supply?

The main limitation with high power in the Kilowatts is being able to source the components.
Transformer, output inductor, mosfets, current sense  elements, they will all be large with few choices of vendors.
You can parallel mosfets as long as you have sufficient gate drivers.
The Transformer will be custom.

The UCC28950 would not have any limitations in itself, the controller really does not know the power level it is commanding.

With the large power components the PCB layout and parts placement will be more difficult.
Longer traces will likely result due to the space needed to accomodate the large power devices.

Usually it is recommended to use LOAD SHARE, with a device like the UCC29002, to parallel multiple power stages in the 2-3KW area.
Finding components at the lower power is easier, layout will be easier.
However, paralleling power stages does not work well with Synchronous Rectifiers.

So if synchronous rectification is desired then you can attempt to scale up a UCC28950 design.
TI has no references for designs in those power levels.

If you can forego synchronous rectifiers then consider creating several power modules and paralleling them with load share.

Dear Ed,

I'm not getting put in share 2 power 2KW sync with the secondary (UCC28950).
I can not share the work because I'm working 50V / 20A output.
I have questions in the calculation of the resistors to set the gain due to high voltage (50V).
Please help me.

thank you

Leonardo Souza
lssouza@linear.com.br

Hello Mr Souza,
Are you using the recommended high voltage circuit for high voltage outputs, in the data sheet.
The base of the NPN is at a fixed DC level, so Iadj from pin 5 will be equal to the current in the collector that pulls on the RADJ.

Forget about sync rectification at 80V. at this voltage with that power levels the only solution is FREDs.

24v is another story but still sync rectification is expensive.

a 150v schottky diode has Vf of around 0.8V and for 250A application you need 0.8/250=3.2mohm at high temperature for each mosfet to be equal with diodes.

So you need a 1.6mOhm mosfet to halving the rectification  loss and it means 2x 0.8mohm/200V mosfet at 25C that will cost a lot.(if the cost matters)

I don't get your second comment.

I agree with NOT using Synchronous Rectification in this case.
1. the high output voltage makes the losses of a diode minimal, as you observed.
2. but more importantly, this design will have modules paralleled and use UCC29002 for share balancing.
    Power Modules with Synchronous Rectifiers do not work well when paralleled.

In Paralleled power designs, one module will always want to be the Master because it's output voltage slightly higher naturally, due to many factors.
See the attached Unitrode Seminar paper on Load Share for Paralleled Power
4863.slup207 paralleling power.pdf

The other parallel modules will want to put out a lower voltage, UCC29002 load share control attempts to force their outputs higher so they can contribute to the load.
If Synchronous Rectifiers are used, the lower voltage modules (slaves) will try to sink current from the master, since the master Vout is higher.
This will cause an extreme degradation of efficiency.

You could use ORING diodes between the modues to prevent backflow of current.
Or see the attached SLUA550 app note using a TPS2412 ORING controller to prevent current backflow.
0726.ucc39002 with tps2412 oring controller slua550.pdf

But the fact remains that modules with Synchronous Rectification must be handled very carefully.

Can you comment on the "load sharing" case and the potential problems you outlined in your post in a dual phase system, ie where two stages operating out of phase, each with synchronous output stages, combine to a total current through their output inductors?

The UCC28950 is intended to easily allow dual phase operation, isn't it? How is this different to parallel operation? Or is it the fact that they are anti-phase synchronized significant somehow?

Dual or Multi Phase is commonly referred to in NON ISOLATED boost or buck converters.
For isolated converters to achieve similar results it is referred to it as Paralleling.

The Phase Shift Full Bridge in itself can be considered a dual phase forward converter, similar to push pull and half bridge.
Each channel is operating 180 out of phase with the other resulting in a net duty cycle approaching 100%.
UCC2895 or 28950 is not intended for non isolated applications,

Forcing current sharing in non isolated multiphase is easier than doing the same for isolated converters.
In non isolated you have easy access to phase currents that can be used by a single controller or routed to mutiple controllers.
With proper use of the control IC, synchronous buck or boost is relatively problem free.

Isolated converters have many more variables, per the previously attached seminar paper on load share.
And their control loop is generally closed in the isolated secondary.
So to force load share you measure the output current and force a lesser slave up to the current the master is putting out.

Also in offline converters at high power there is a PFC stage followed by an isolated stage.
The isolated output is the best place to attempt parallel sharing with these multiple stages of power conversion.
Each stage is more of a MODULAR approach as opposed to a multiphase DCDC buck.

Synchronous rectification at high KW power levels may have problems with paralleled converters.
When hot plugging a module the new module takes some time to come up, soft start.
During that time it could actually be sinking current from the master module.

Or, under severe load transient conditions it has been noted that reverse current can flow in the synchronous rectifier stage causing very high transient voltages.

There are many more considerations when paralleling isolated converters than when designing multiphase buck or boost.