Coupled inductor SEPIC is wrong?

[eem2am]

Hello,

Why does Dr Ray Ridley, one of the world’s leading SMPS Engineers, state that the SEPIC converter with coupled inductor should never be done?

The last paragraph, on the right hand side of the third page (actually called page18) of the following states that the SEPIC with coupled inductor should never be used.

http://www.switchingpowermagazine.com/downloads/Sepic Analysis.pdf


So what’s wrong with the SEPIC converter with coupled inductor?

 

[BobK]

That paragraph is self-expanatory. A SEPIC converter has non-coupled inductors, if you couple them, it is no longer the same design.

Bob

 

[eem2am]

SEPIC with separate inductors is a coupled inductor SEPIC with k=0.

Coupled inductor sepic is a well konwn and accepted topology.

-please see coilcraft's range of coupled inductors for sepic converters.

Non coupled sepic gives great problems with ringing between the L's and C's, especially if the crossover freq is not at least three times the 2L,C resonant frequency.

 

[(*steve*)]

You have two apparantly authoitive sources giving conflicting advice for a circuit that is considered (as one of them says) "poorly understood".

The one recommending non-coupled inductors is in a publication featuring lots of information about modelling the circuit.

The other information comes from a manufacturer of inductors (and should be considered somewhat qualified to give advice on their use).

I think there are a number of possibilities:

a) one or both are simply wrong. I could not judge this to be the case as their analysis is likely to be way more complex than I could understand, and thus beyond my ability to critique.

b) The circuit is more complex than either have suggested, and both have it correct within a particular set of constraints.

c) some other option that I can't come up with because I'm simply not clever enough.

I'd be leaning toward (b) or (c) and my first 2 options.

If you can go through the design process for both, and you find that each produce a non-overlapping solution (e.g. where each model suggests the other one is poor or unstable) then the issue is more likely to be (b)

Once you understand both of them fully, maybe you can come up with the Grand Unified Model of SEPICs.

If you do, you will forever be known as "Mr GUMS"