Think of a circuit model that lets you move magnetic leakage around like sliders on a synth, without changing the external behavior of your coupled inductors. [Sam Ben-Yaakov] walks you through just that in his video ‘Versatile Coupled Inductor Circuit Model and Examples of Its Use’.
The core idea is as follows. Coupled inductors can be modeled in dozens of ways, but this one adds a twist: a tunable parameter 𝑥 between k and 1 (where k is the coupling coefficient). This fourth degree of freedom doesn’t change L₁, L₂ or mutual inductance M (they remain invariant) but it lets you shuffle leakage where you want it, giving practical flexibility in designing or simulating transformers, converters, or filters with asymmetric behavior.
If you need leakage on one side only, set 𝑥=k. Prefer symmetrical split? Set 𝑥=1. It’s like parametric EQ, but magnetic. And: the maths holds up. As [Sam Ben-Yaakov] derives and confirms that for any 𝑥 in the range, external characteristics remain identical.
It’s especially useful when testing edge cases, or explaining inductive quirks that don’t behave quite like ideal transformers should. A good model to stash in your toolbox.
As we’ve seen previously, [Sam Ben-Yaakov] is at home when it comes to concepts that need tinkering, trial and error, and a dash of visuals to convey.
Doing math may take literal weeks. Redesigning a circuit to fix the problem will usually take 1 or 2 hours with a soldering iron and a scope. If it’s not done for a hobby I don’t see it as viable option – that is unless you’re actively looking to get fired for “doing engineering”.
You are right (although weeks is an exaggeration), but if you’re designing something professionally, you should do both. Fix it quickly with the scope to buy time, then do the math to know why it works
A soldering iron is not what you use for designing anything.
I’ve designed switchmode converters with a very high power density (for an aircraft). The design takes months before you build anything, only than you build a prototype, measure and probe, and iterate until it’s perfect.