Synchronous rectifier drive chip is unacceptable?

[cupoftea]

Hi,
The attached shows a Full Bridge with Synchronous Rectifiers. (LTspice and PNG)
48Vin, 12Vout 200Wout, 117kHz.

At 3ms in the LTspice sim, the load gets suddenly switched out from full load to no load. As the inductor current falls,
it then reverses, -whilst it is in reverse in the synchronous FET, the synch FET is then turned OFF. This is "breaking an
inductive current"...not just some little stray inductive current, but the inductive current in the power inductor!

These overvoltages are so high that they overvoltage the synch FETs and their paralleled diodes.

It may be said that the LTC3901 offers reverse current sensing in the synchronous FETs, and that it would switch them off
before this current got too high. However, this is not acceptable. The signal from any sense resistor would be very small, and
drowned in noise at low levels of reversing current.
In any case, by the time the power inductor current has actually gone into reverse, its way too late to switch off the sync rect FETs.

May i ask Why has this chip (LTC3901) been manufactured?

It is unsuitable for purpose. It doesnt even have a shutdown pin
so that the synch rects can be turned off as the inductor current falls towards zero.

 

[Easy peasy]

are you making a whole lot of assumptions based on an imperfect simulation model ?

 

[cupoftea]

Thanks, i think the principle that "detecting reversing current in a synchronous fet" is wrong.
And this chip does that very thing.
The inductor current should never be allowed to reverse in any synchronous FET.

......it would be easy to arrange this.
Eg turn off the sync rect fets before the inductor current goes down to zero

..Other way is to switch the fets all the time so that when inductor current is net zero, its actually shuffling back and forth....and when a synch fet switchs off, the inductor current is going through its diode at that time....so no breaking of an inductive current.

 

[cupoftea]

Though thanks, its almost unbelievable that no semico on the planet has any kind of offering for this ulta-common requirement.
Do you know why this application area is such a techno-void?

I mean, a synchronous rectifier is just a FET which switchs on and shorts the diode out...simple as...but nobody does anything half decent for it for isolated SMPS.
Sure we can do a DIY but the component count goes right up....because you need the delay blocks, and the current sensor and comparator to disable it all when current goes toward DCM. Nobody even does any "offtheshelf" module of this.

Must admit, 10 years ago i rev eng'd Europes best offering for 3kW EV charger....and it had NO SYNCH RECTS!!!...just to220 diodes stuck to casing...and needed a water cooling jacket!!
Every other PSU that i rev eng'd also has no synch rects....mind you thats partly because the leakage L resonates with the CDS of the synch FET..and that simply cannot be damped out as it would mean too much dissipation.