Use of a diode with "snappy" recovery behaviour

Sometimes its better to use a slow diode with plenty of storage charge to reduce the rf interference caused by a very fast diode switching off...

I've had cases of EMC failure because even in a mains linear power supply the designer had specified fast diodes, replacing them with slow (normal) diodes reduced the rf emmissions enough to let the instrument pass the EMC testing.

John.

What he speaks of is the fact that a true square wave is the sum to infinity of all the odd order harmonics of the fundametal envelope frequecy. If the edge of the square wace is sharp enough to cut paper you can detect RF noise to hundreds of Mhz and it interferes with 2 way radios etc because at these frequencies the radiate easily from short conductors. If slower diodes are used these edges are rounded and the frequency distribution is far far lower and the total energy in the harmonics that are radiated is far less. There are other means used to suppress this, ferrite chokes and capacitors etc.

Since the driving force is efficiency and the faster diodes are more efficient, a tradeoff ensues.

You may want to explore syncronous rectification. Many vendors of power supply chips are offering this now in their designs.

Yes, synchronous rectification uses FETs that are turned on and off to rectify.

They help in efficiency as they only have a small IR drop instead of the common single or dual diode drop in diode based rectification. In 5 volt circuits the diode drop can cost you almost 10% in efficiency

What is so amazing (to me, at least) is that synchronous rectification, as a technique, has been around for decades and that we're only now seeing this technique incorporated into more general designs. It's not that the parts required are so exotic or expensive that advances in technology have made it only now affordable. Maybe more seasoned power-supply designers here can offer an insight as to the why? I'm stumped.

-e

In smaller supplies I think the cost to benifit ratio did not justify it. Sometimes in larger supplies the configuration does not easily lend itself to a working configuration. But as with all things new ideas have driven the state of the art to a point where it is more easily done and less expensive to do. Also, people tend to stay with what they know until it hurts enough to change.

Thanks, rcapper.

On a different note, I had a discussion at work today about who authored your sig's quote. B.R. did, of course. Thanks!

-e

Synchronous rectification-on-a-chip

http://www.google.ca/search?hl=en&q=%22synchronous+rectification%22+%2Bchip&btnG=Google+Search

Back in the old days when FET's didn't have built-in fast-recovery rectifiers, I had very good success in eliminating fly-back voltages which caused self-commutation of half or full bridge amplifier transistors. Wasn't long before "ultra-fast recovery" rectifiers were on the market. Schottky was popular because of the low forward voltage drop, thereby minimizing the losses. Do a search on fast or ultra-fast recovery rectifiers.

Thanks all. I well know lacks of diodes with "snappy" recovery behaviour. But I am interested more with their advantages. I think, that the high-voltage "snappy" diode with correct snubber circuit will have better characteristics, than fast or superfast diodes in 1-5kV region. Therefore I repeat my question: Does anybody have positive experience of use of the high-voltage fast silicon diodes (600V - 3kV) with "snappy" recovery behaviour in SMPS with "hard" switching (buck or boost converter for example)? Krass

Hi,

I have got interest in same topic you asked for. Have you made any progress in meantime within this topic, or could you guide on available diodes with suitable reverse recovery charge and additional snappy recovery? Im looking for diodes within 600-1kV region for improvement within boost stages.

buau