800+V MOSFET with fast Antiparellel diode

Why would you use a full bridge at that power level? Is this a typo?

RL

Yes we are using a full bridge at this power level because that is how the design was originally made. The MOSFETs used in the original design are hard to get (long lead time) and I'm looking for suitable replacements.

Is there anything wrong with using full bridge for lower power levels with high voltage?

Greetings MJ..

A side note.

Reference: "The application is a 150Khz ..."

Megahertz is defined as MHz, kilohertz as kHz and hertz as Hz. It is, in fact, ALWAYS a capital "H" to pay homage to Mr. Heinrich Hertz. The first letter of that last name is always capitalized. Consider:

or, or, (click on "M" or "J-K" - these folks should know the difference)

Further proof? Take a look at

and note their frequency references. In addition, simply take a look at a stereo dial, clock radio or even your transistor radio and notice how the manufacturers abbreviate frequency.

Cheers, Mr. Mentor

600VDC at | ~150Watts. | | I can live with highish RDS (1 - 8 ohms) as long as the part is fast | switching (lowish ciss) and fast trr. | I'm trying to aviod external antiparallel diodes across each fet. | | I'm familiar with the fets from the following sources and am looking for | more options. | | Vishay IR | IR/Omniel | APT | ST | Fairchild | Infinion | IXYS | Onsemi (nothingover 600V) | | Are there any other manufacturers I should be looking into for fast | switching high voltage mosfets? | | Any IGBTs that can have good efficiency at 150KHz could be considered as | well. | | |

Well, you can cut complexity and many other factors in half simply by converting to a half bridge, for a start, ( particularly if the output voltage is appreciably lower than the input - which I suspect is the case ).

Reverse recovery problems and leakage energy recovery can be simplified or avoided entirely by using a two-transistor forward ( or even two-transistor flyback ) converter.

Justification for the original design is really more in order here, particularly if it's producing problems that seem to require redesign, now.

RL

email by removing 'nospam.' .

I'd love nothing more than to chnage the design (design is the fun stage) but this is one of those fire that poped up and they want a replacement component pronto. The FET in question is an APT1003RKLLG. nonstock te digikey and all microsemi parts are on do not design in status due to delivery problems.

There's no reason to use parts higher than 800V fets in a full bridge with 600V rails, but

TO-220 STP3NK80Z STP4NK80Z STP9NK80Z STP11NM80 STP8NK85Z STP2NK90Z STP5NK90Z STP6NK90Z STP9NK90Z STP5NK100Z STP8NK100Z STP3N150 STP4N150

TO-220 insulated STP8NK80ZFP STP10NK80ZFP STP3NK90ZFP

are all in stock at Digikey

RL

The original part was 500nS/3.2uC

Placing external antiparallel diodes may not be effective without series drain diodes to stop parasitic diodes from functioning. I'm surprised that you are running into this problem at such low current levels.... should be a few hundred milliamps.....unless you're really flogging the mag current.

It may not be such a big deal converting a full bridge to a half bridge or two-transistor forward - its possible that the foil patern would not have to change - just the magnetics and parts stuffed in present locations.

Full bridge to half bridge - replace one side of the bridge with film capacitors, halve the primary turns and lay off current mode in favor of voltage mode. You can still use current mode influence for ripple and transient rejection. A little goes a long way. Same 'reverse recovery' in the fets, though.

Full bridge to two-transistor forward. Replace one phase of fets (opposite quadrants) with ultrafast rectifiers, reroute secondary center-tap to antiphase end of secondary, increase secondary rectifier voltages, and output choke inductance increases by ~3. Reverse recovery is all in real rectifiers. The old control circuit will limit switching duty cycle to

Uhm, I thought he founded some kind of car rental business.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.