Coupled inductor for multiphase buck converters

[bhl777]

Hi All, I found some applications note that using coupled inductor in multiphase buck converters can help in both efficiency and transient response. It seems like it is good enough to replace the conventional inductors in this application.

However, there are many people still use conventional inductors. Are there any shortcomings of using coupled inductor in the performance wise (in additional to the cost)? Thank you!

 

[FvM]

I don't see how it's supposed to work. Please refer to application note.

In any case the coupling factor must not be too high.

 

[bhl777]

I don't see how it's supposed to work. Please refer to application note.

In any case the coupling factor must not be too high.

Hi FvM, this is one example: https://pdfserv.maximintegrated.com/en/an/TUT5997.pdf , thank you!

 

[FvM]

Thanks, I was imagining inductors with positive coupling coefficient similar to common mode chokes. But these inductors implement negative coupling, a kind of multiple interphase transformers.

Obviously the magnetic circuit design is challenging for more than two or three phases.

 

[bhl777]

Thanks, I was imagining inductors with positive coupling coefficient similar to common mode chokes. But these inductors implement negative coupling, a kind of multiple interphase transformers.

Obviously the magnetic circuit design is challenging for more than two or three phases.

Hi FvM, people use one coupled inductor for each 2 phases in a multi-phase system (for example, a 6-phase system will require 3 coupled-inductors). From the pespective of the user (not the magnetic circuit designer), are there any potential problems in using these kinds of coupled inductors?

Thank you!

 

[treez]

Interesting, the thing with coupling inductors of paralleled bucks is that they say in your article that it helps to be conducive towards faster transient response of the overall paralleled buck converter.
The thing is, the paralleled bucks with coupled inductors will have to be exactly in phase with each other’s switching. That means they all switch at the same switching frequency aswell. –If you avoid coupling the buck inductors, and instead use multiphase interleaving…then your “effective” switching frequency goes up to (n * buck Fsw), where n is the total number of bucks. When you have such a high effective switching frequency, you find that this more easily facilitates having a faster transient response also….but you avoid all the cafuffle of having to get your inductors coupled, which is inevitably a custom wind.
So I would say that the coupled buck solution is not necessarily all good. Also, multiphase interleaving also results in a much lower current ripple, -and this as you know, was said to be one of the advantages of using couple inductors.
One unmentioned advantage of having paralleled bucks with coupled inductors however, is that in some cases, it is seen that one of the bucks (if they are synchronous) starts having its current going backwards…….ie, it starts acting as a synchronous boost. This can sometimes be due to infringement of a minimum on time in one of the bucks….(M.O.T has a tolerance in each of the controllers). If you have coupled inductor bucks, then you reduce the chances of this happening.

 

[FvM]

The thing is, the paralleled bucks with coupled inductors will have to be exactly in phase with each other’s switching.

No. The discussion, also in the linked literature, is about interleaved multiphase buck.

Hi FvM, people use one coupled inductor for each 2 phases in a multi-phase system (for example, a 6-phase system will require 3 coupled-inductors).

That's a possible solution, with 180 degree phase shift between both phases of each pair, a classical interphase transformer topology. But it's not fully utilizing possible advantage of coupled inductors as addressed in literature.

From the pespective of the user (not the magnetic circuit designer), are there any potential problems in using these kinds of coupled inductors?

I see no problem, but not necessarily an advantage, It depends on inductor parameters. You should calculate and simulate a practical design to evaluate possible advantages.

 

[schmitt trigger]

Many years ago, in the mid-1990s, I went to a very interesting power electronics seminar by Gordon "Ed" Bloom, specifically about integrated magnetics.

Integrated magnetics can achieve significant performance advantages, but their design and analysis are more difficult.

Also, the actual manufacturing of the magnetic devices is more difficult, and all of its parasitics have to be managed closely.

All in all, it is more challenging, and therefore -in my personal opinion- designers shy away from them if the performance requirements are not needed.