how the laser happened

My switching power supply filters are usually dominated by the first inductor. It has to let some tolerable ripple current into the downstream caps, has to not saturate, and must not get too hot in the minimum expected air stream, from core loss and copper loss. And fit available space and not cost too much and be available for purchase.

I'll often have a secondary high-current ferrite bead to reduce EMI spikes, typically maybe a per cent of the main inductance.

None of that is classic filter theory.

Only Spice can predict the power supply load response. It's too nonlinear for classic filter theory.

There are cheap tricks to compensate the control loop, once the big power stuff is designed.

Yes, this is exactly how all the radar power engineers of my acquaintance solve the problem. LT Spice is their standard tool.

Yep.

Joe Gwinn

It certainly helps to know some control theory and classic filter theory, but that's just a guide to instinctive design and loop tuning.

Current limiting further complicates dynamics. More cases to simulate.

Some topologies get very different at light loads, when they go discontinuous. I avoid them whenever I can.

Spice-Tweaking filters beyond 3rd order is hard. It's easy to diverge, to get lost in space. One trick there is to take the AC feedback early, close to the switcher, before a zillion phase lags pile up. Starting at the switch node is cool; the transfer function between PWM demand and voltage there is a dimensionless gain.