Coupled inductors for filtering

What are the tradoffs to consider when deciding whether to use coupled differential mode, coupled common mode or discrete inductors for filtering in AC applications.

For example below is the diagram of the winning google little box challenge (DC to AC inverter). All inductors are coupled. On the inverter output why chose to have a common mode followed by a differential mode inductor instead of two discrete uncoupled inductors (which will also block both differential mode and common mode).

Ok I can give an obvious answer: It allows independent tuning and optimizing of the common-mode versus differential mode properties. That makes sense but as I survey parts I see discrete single inductors in more highly optimized packages (like molded). In a case I'm studying now its smaller to have two single molded inductors than a wound common mode choke of similar value and rating. Why not go that way and get another differential stage for free?

Hi,

a common mode choke may be very high impedance for common mode signals, but it may be very low impedance for differential mode signals.
You can get the same when you use two not coupled chokes.

In one case the inductances add to twice it´s value, in the other case they subtract to almost zero.

Klaus

I see discrete single inductors in more highly optimized packages (like molded). In a case I'm studying now its smaller to have two single molded inductors than a wound common mode choke of similar value and rating.

Click to expand...

Is this also the case when you consider load current and inductor saturation limit? A typical common mode choke rated for e.g. 6 A with mH inductance can't be implemented at same size with two individual chokes.

Ok I may have made a terrible choice of common mode choke for comparison:
https://www.coilcraft.com/msd1260_cm.cfm
12x12x6mm/2.5A is only 10uH

This part can be easily matched with individual inductors. But this family is also marketed as coupled inductors (like for SEPIC applications) and may have saturation currents comparatively higher than typical "common mode chokes" I guess?


By contrast this line has similar sized parts with ~100X the inductance.
13x13x5mm/2.6A/1mH
https://www.coilcraft.com/sm_pl_filter.cfm

That makes more sense.

But back to theory: Common mode chokes are able to have much higher impedance to common mode currents because the core 'ignores' much larger differential currents. But am I correct in saying that differential mode coupled inductors have little theoretical benefit in cases when differential mode currents dominate? You don't get a smaller core by 'ignoring' tiny common mode currents. So the choice of individual chokes or coupled differential-mode choke is a practical one?

Hi,

Generally there are (at least) two types of inductors:
* EMI / EMC filtering inductors. Low Q for damping unwanted frequencies. Often high inductance.
* power inductors, high Q for low loss.
Sometimes they look identically, easy to confuse.

Choose the right type according application.

Klaus

But am I correct in saying that differential mode coupled inductors have little theoretical benefit in cases when differential mode currents dominate? You don't get a smaller core by 'ignoring' tiny common mode currents. So the choice of individual chokes or coupled differential-mode choke is a practical one?

Click to expand...

Yes. I'm not even aware of off-the-shelf common mode differential chokes.

- - - Updated - - -

By the way. I think, the most interesting point of this award winning design is the role of coupled inductors L1/l2 and L4/L5 and phase shifted half bridges in achieving "soft switching for the entire operation range".

I've read the paper a few times. Do you think you understand exactly what they're doing? The most they say is:

"digital
control based on a fast microcontroller combined with a CPLD; fast measurement of input/output
currents and voltages; efficient feedback on the switching events of the HBs; a learning algorithm
for the active filter; optimization of the switching frequency between 35 and 240 kHz depending
on the output current; a variable phase shift between the HBs (0° or 90°) and a dead time
modulation of the five HBs (50 ns to 3 µs). "

So just two different phases. Otherwise the wide frequency and wide dead time ranges speak to standard half bridge ZVS schemes in my mind (allow huge ripple so ripple currents change sign every cycle). Also note the 5th active filter leg has an individual choke but is also stated to ZVS. So I've assumed that phase shift and coupling plays a more minor optimization role.