LTC3901 synchronous rectification in a 2.5kW PSFB not working correctly

[Ashwin619]

Hi,
I am planning to design a 2.5kW (84V, 30A) PSFB converter (offline with Vin = 390V) for a battery charging solution. I have somewhat understood the ways by which this topology achieves ZVS and the significance of Leakage/shim inductance, Mosfet Coss, and deadtime in order to achieve this.

I am currently simulating this above application using LTC3722-1 and LTC3901. The values of the passive components like timing resistor, voltage dividers for fixed and adaptive delay programming, current sense, SBUS, UVLO, and timer (ltc3901), etc have been done as per the calculations in the datasheet. Some of the other values (like filters and compensation) have been referenced from typical application design.

I have tried to design the transformer (in the simulation) considering that Dmax ~ 70% and Dtyp ~ 65%. This yields a turn's ratio (Np/Ns) close to 3/1. The inductor ripple is set to 20% of the output current (i.e. 30A). I have got a magnetizing inductance value of 1.37 mH using the above parameters for an Fs of 100 kHz. I have used the calculation note SLUC222D (attached below as UCC28950 2.5kW excel file. Originally I had intended to use the TI UCC28950 for this design however its currently unavailable along with a ridiculous lead time)

The simulation along with it's feedback compensation is working somewhat fine, however, the gate pulses of the LTC3901 don't look anywhere as close as expected (i.e. very short pulses). The output inductor (47uH) seems to go in DCM mode which I had assumed to be the main source of error of shortened gate pulses, as the ME+ may rise above ME- causing the termination of pulses (same for MF+ and MF-). The SR mosfet dissipation is very high at almost 180 - 200 W during this scenario.

After changing the inductor value to around 111 uH and output cap to around 2000uF the DCM mode ceased to exist although there didn't seem to be any changes in the gate pulses. The mosfet dissipation improved somewhat to around 70 - 80W which is still quite high.

I feel like this case of high dissipation is mostly caused by the flow of forwarding current through body diode rather than the main channel of mosfet (due to gate turn off) or either reverse flow of current through body diode.

Note: I have changed the value of the sense resistor to 1m Ohm as that of the recommended value of 21.8 m Ohm considering Iprimary p-p to be around 13.59A. This was done as the IC was terminating the gate pulses perhaps due to the initial inrush current. I guess changing the lead edge blanking resistor RLEB would solve this issue.

I am attaching the simulation for both the adaptive and the fixed delay one. I would really appreciate it if someone could shed some more light on this issue. Any help or suggestion is appreciated.

Regards,
Ashwin619.

[Ashwin619]

I have noticed that the diagonal Mosfets A and D almost completely overlap each other (same for C and D), also the synchronous rectifier pulses from LTC3722-1 seems to be affected by this as there's hardly any freewheeling period in the secondary side (the period when both the Mosfets are on and current flows through both of them).

I also fear that having lowered the value of Rsense might cause some issues as the IC being a Current mode control must be getting its ramp signal from the current.

Currently I am referring to the datasheet in more detail in order to find a solution for this. let me know if I have missed something.

[CaptDon]

Sounds like you are working on a WABTEC/GE India Rail project?? Has our B.E.L. Battery Electric (Piece of Shit) ooopps I mean Locomotive design now being done in India? If it wasn't for the U.S. government handing out money for stupid ideas no one would try this (again). The blue and green road number 2010 was a hybrid locomotive. Guess what, it isn't anymore. It was converted back to a standard unit. The pure battery electric is even dumber. If a hybrid isn't worth the effort (and it wasn't) a pure battery unit is even more foolish. Anyway, good luck with your simulation and design hope someone here can help you arrive at the answers you need!! Cheers!!