Low Rds MOSFETS - Power Rectification

[iceblu3710]

I found an interesting HEXFET at the market and just started poking around on digikey and apparently I missed something in the world of semiconductors!

I had no idea their where MOSFETS out there with an Rds(ON) of <5mOhm. That and theirs a strange DirectFet Isometric package I have never seen before. I found some 100V, 180A FETS with a Rds of 3.2mOhm, at full bore that FET only drops 100W of power which is phenomenal.

It got me thinking that I should re-design my high current PSU to use much more efficient parts as right now its driving a 1HP DC motor and its a space heater at only 14A.

The issue I see now is the conversion of AC to DC using a bridge rectifier. with a 1V diode drop at 180A (x4) your rectifier is producing 720W of heat which is ridiculous. If you had the ability to use these new MOSFETS you could cut wasted power down to 400W.

Are their any rectification solutions out there that I missed the boat on or any other ideas to help reduce heat before I get designing?

Thanks!

 

[dick_freebird]

Synchronous rectification has been around for a long time.
The thing is, you don't have a simple forget-about-it diode,
you need a properly phased gate drive (and the drain
current had best not reverse, lighting up the body diode).
It has great advantages, simplicity is not among them.

 

[leo_o2]

MOSFET always have lower voltage drop than any diode in power applications.
However, 5mohm resistance should be very new technology for 100V MOSFET.
Nowadays, some new technologies like cool MOSFET come up.

 

[iceblu3710]

A few interesting MOSFETS:
Digi-Key - IXFK320N17T2-ND (Manufacturer - IXFK320N17T2) 170V, 320A, 5.2mOhm @ 60A
Digi-Key - IRF7749L2TR1PBFCT-ND (Manufacturer - IRF7749L2TR1PBF) 60V, 375A, 1.5mOhm @ 120A


I'm not sure what categorize to look under to find synchronous rectification controllers but they must exist because of the careful timing and dead time requirements so everything doesn't go up in smoke. I found a few self biased SR circuits but they are not as efficient as active control.

I'm sure I could use a uC such as ATMEL's PWM3 series as they have internal dead time generators and are guaranteed not to short on start up but to develop the system I would need to rig up a very current limited AC line source and that's more of a hassle than anything.

 

[walkura]

IRF has synchronous rectifying controllers (IR1166 and IR1167, although there seems to be a new version out) they work very nice.
They also have a whitepaper on how to use them on 50 Hz (http://www.irf.com/technical-info/whitepaper/TP-080527.pdf).
One little note of caution ,do keep a good eye on the tolerance of the components you use, those 12 nanoFarad and 56 kilo Ohm resistors should be reasonable close to the value shown in the schema.
Otherwise you will have heating and probably destruction of your mosfets.
Also have a look at this mosfet, IRFP4468.
I hope this is of use to You.

Good luck and be carefull.
Walkura.

 

[iceblu3710]

@walkura: Thanks for that link! I have seen several uC and self running SR circuits and they where VERY complex and I did not know where to begin but that IR1167 is an awesome IC. Makes the job very simple. I would be using 1% carbon film resistors and 5% polyester caps or even NPO to keep things as stable as possible.

Its a bit more pricy than a simple diode bridge and comes it at about $26 for the FET's and IC's but if I can get it to turn out 140VDC @ 15A I can seriously reduce the size and heat output of my lathe controller.

 

[walkura]

@iceblu3710:
There are more manufacturers with controllers like this, but IRF seemed to be the only one providing a white paper on how to use it with 50 Hz (when i been working at this about 18 months ago that is)
With 1% resistors and 5% caps it should work like a charm, but keep in mind this is designed for 50Hz and if You live in a country with 60Hz You might have to "tune" those resistors and capacitors to your frequency.
If You have this "untuned" for instance a buffer elco might cause large "back" currents (your duty cycle might be a nice 50% regardless of the state of charge of your bufferelco)
Thats why those tight resistor and cap value's actually matter.
Also it might be cheaper to have 2 smaller mosfets parallel, those IR1166 can provide upto 7 Amps of gatedrive and sometimes 2xTO220 can be cheaper for the application then one large die mosfet.
You just have to look a bit around to what mosfet will fit best (or better said be most economical) for this application.

Have a nice eve and goodluck with your project.
Walkura.