Synchronous Buck Converter Switching

[sodoyle]

TL;DR Summary

Are synchronous buck converters soft switched?

For a synchronous buck converter, does soft switching have to be implemented? I assume there will be some ringing if not. If soft switching is required/recommended, can it be implemented on just one of the switches to simplify control and still have most of the benefit?

 

[berkeman]

I've never used any "soft switching" in my buck converters; is it a common thing? Or is there something special about the synchronous buck design that leads to that need? I've only used bucks with flywheel diodes, not a synchronous lowside switch...

Do you have any links to discussions about "soft switching" in DC-DC converters?

 

[sodoyle]

It's just possible to get a higher efficiency with synchronous. Of course that comes at the cost of increased control complexity and an extra FET so it's really just a matter of playing the tradeoff game.

Here's a few open access paper discussing soft switching in synchronous converters
https://jpels.org/digital-library/manuscript/file/16893/9_JPE-15-04-044.pdf

https://www.hindawi.com/journals/ape/2008/862510/

https://www.researchgate.net/publication/26530476_A_Novel_Soft-Switching_Synchronous_Buck_Converter_for_Portable_Applications

 

[Baluncore]

I believe a simple flyback diode would be classed as “soft commutation”. If you parallel the flyback diode with a MOSFET you can improve efficiency by reducing the flyback diode voltage drop. The gate of the parallel MOSFET would need a control signal from somewhere, and could then be called “synchronous rectification” or even “hard commutation” because it is forced.

 

[sodoyle]

I believe a simple flyback diode would be classed as “soft commutation”. If you parallel the flyback diode with a MOSFET you can improve efficiency by reducing the flyback diode voltage drop. The gate of the parallel MOSFET would need a control signal from somewhere, and could then be called “synchronous rectification” or even “hard commutation” because it is forced.

I've never heard of it as "soft commutation" just because of the diode. The anti-parallel diode is used in a lot converters with "soft commutation" though. That's largely because the schottky diode is often a better diode than the FET's body diode. Some type of diode is needed, body or anti-parallel, to implement deadtime which ensures both switches aren't on at the same time during commutation. The external diode is a way to reduce dead time losses.

 

[Baluncore]

If a switch such as a diode turns on or off through it's inherent characteristics and position in the circuit, then it is "soft commutation". If it switches because it is controlled by an external signal then it is "hard commutation". The anti-parallel diode, or the body diode across a mosfet is inherent = soft commutation, the mosfet is controlled = hard commutation. If you replace the diode with a controlled synchronous rectifier, then it becomes hard commutation for both the on and off transitions.

Y. Chéron (auth.), C. Goodman (eds.) - Soft Commutation - Springer Netherlands (1992)

4.1 INTRODUCTION
In the Chapter 3, it has been shown that AC supply of inductive or capacitive loads can be achieved very satisfactorily (high efficiency, low distortion). It has also been shown that the performance of these resonant inverters was tightly linked to the commutation process in these switching devices all of which execute one controlled switching and one inherent switching.
We will define as a soft-commutated direct converter a converter only involving switches, each of these switches having at most one controlled switching and at least one inherent switching, the controlled and inherent switchings being if necessary different from one switch to another. So the soft commutation is not compatible with totally controlled switches. In soft-commutated converters the power transfer is controlled by means of switches or by a set of switches with thyristor or dual thyristor switching features.
The term 'soft commutation' is legitimate since those switches that have one controlled switching and one inherent switching can be fitted with lossless snubbers (a series inductor when turn on is controlled or a parallel capacitor when turn off is controlled) so that the switching losses are considerably reduced.