When cascading a ferrite bead and inductor in the form C-L-C-L-C, which is best to put first, particularly for a buck converter?

Question

I am asking this question specifically for buck converter inputs, but am happy to consider answers that are more general -- not specific to buck converter ICs, or focused on the output filter.

Background:

The datasheet of buck converter IC MPQ4425C claims to pass CISPR25 radiated and conducted emissions with a particular EMI filter.

A snippet of the MPQ4425C datasheet which contains the recommended EMI filter, is pasted below. The datasheet can currently be accessed at the following link: https://www.monolithicpower.com/en/documentview/productdocument/index/version/2/document_type/Datasheet/lang/en/sku/MPQ4425CGQBE-AEC1/document_id/10541/

My framing and assumptions follow:

This EMI filter contains many of the design decisions that I would consider necessary for good buck converter EMI performance. Buck converters have discontinuous input current (square waves of current), so input filtering is important. It has parallel MLCCs on the input, and at least one series attenuating element (here, both inductor and bead), with capacitors on either side of the L for further attenuation. While not strictly true, a device like this is likely to have input wiring that is able to radiate, further necessitating attenuation of voltages/currents on said input nodes.
Buck converters in general have much lower frequency content on the output (triangle waves of current), and as such the output filters and capacitors are less of a threat of EMI.

Thus, I intend to focus on the input of this circuit, but am open to re-framing my assumptions. Ultimately I hope for answers/discussions that get to the crux of the question, which is the decision to place the ferrite bead or the inductor first.

In short:

Cascading LC filters is a way to increase attenuation. Combining two different kinds of inductors, a standard inductor and a ferrite bead, is another way to improve frequency specific attenuation.

But in which order should a ferrite bead and an inductor be placed? What effects might each configuration have -- both functional and relating to EMC? In other words -- what if I swapped FB1 and L1?

Edit for clarity: Example of buck converter's square-like input current waveform: Observe square-like discontinuity in input current waveforms. Source: Rohm website about Input capacitor selection https://techweb.rohm.com/product/power-ic/dcdc/2452/

Answer 1

For the input filter, it's important to have the capacitor close to the supply terminal, with a small loop area to GND. This cap supplies the high-side transient.

This would imply C-L-C-L-C input sequence as shown in the MPQ appnote to be preferred.

Layout matters too: minimize the loop areas for both the main input cap and especially the output cap. The fast currents in these paths have a profound influence on radiated emissions.

Here's a sim to play with (simulate it here):

You can see that the current ripple at the 5V input is pretty well knocked down by the double-pi filter.

But also take note of transient current going to the FET drain. This has that square/sawtooth shape as the FET switches on. At first it jumps up immediately, then ramps up with the inductor current build-up. That initial step is a pretty big dI/dt jump, that's handled by the 10uF cap that's closest to the FET. That cap's performance would be far less effective if it were placed with the inductor in its path.

You may be curious how to model a ferrite. The simple model uses 4 components, arranged as below:

From https://www.analog.com/en/resources/analog-dialogue/articles/ferrite-beads-demystified.html

In this article they analyze a selected bead and create a model for it, and show work to optimize its performance with damping.