Switch Mode Converter Controller IC - Choice of MOSFETs

[frasdoge]

Hello,

I'm taking a look at building a switch mode converter for 120V DC input, 5V output. I've come across a controller IC that looks appropriate. LTC3895. Something I don't understand however is that they use different N channel MOSFETs between their MTOP-MBOTTOM pair shown in the example schematics using BSC520N15NS3G and MNDRV using IPD320N20N3.

I've taken a look at the datasheets and can't really see what difference compels them to use different parts here. Those parts of obsolete now but their replacements do not differ in price by almost anything either. Can anyone spot what the rationale is?

Mostly just a curiosity as I plan to swap out the MTOP and MBOTTOM for much lower Rdon, and MNDRV is unnecessary as it's only needed for voltage following at low input voltages nearing the MOSFET VGS, which is not applicable to my application.

Cheers

[David Hess]

Conduction losses are proportional to duty cycle and highest efficiency is about where conduction and switching losses are equal.  So if the duty cycle is very low, as would be the case with a conversion from 120 to 5 volts, then the optimum MOSFET size for the high side will be smaller.

[youngda9]

David's answer is correct.  Also look at some other parameters.

Top FET is 1/2 the price, and 2 are used in that location.

RDS-on is a bit less than 2x the lower FET, so effective resistance will be lower for the two top FETs, and each will handle 1/2 the current, ideally, so switching losses will be reduced.

Upper FET is <1/2 the capacitance of the lower FET, so switching losses will be reduced as well using 2x in the upper location as opposed to one of the lower FET types.

Net results is better efficiency in this setup.

Lower FET will conduct about 95% of the time.  Went with a more costly lower RDS-on FET in that location.  Could parallel FETs for better efficiency, but it would cost more.

These are the trade-offs they chose. 

I've done similar designs recently. 
 - LTC3891EFE based 24V to 12V converter @ 10A output.  Used a single upper and lower FET since duty-cycle will be 50%.
 - LTC3892EFE based 24V to 20V converter @ 20A output.  Used dual upper and a single lower FET since duty-cycle will be 83%.  This setup resulted in more optimal efficiency.

[T3sl4co1l]

I'd just go with a flyback.  Forward or resonant if higher power (>50W maybe?).  Direct bucking at that ratio isn't very effective.

Also gets more headroom for transient protection; 150V rating for 120V (nominal?) doesn't exactly instill confidence.  Or is that the peak including transients already?

Tim