Shielding Inductors

Yup, that's about my rule of thumb. Greatly depends on the kind of inductor. Core? No core? Toroid?

Rotating inductors 90 degrees to each other greatly helps in reducing the coupling.

Jittery DC, as some people say :-)

Hey, are you defecting from the world of IC design and entering board level turf?

Okay. I give up. I can't resist it:

And they say it is feasible, too. ;)

Jon

Like all such rules of thumb, that was developed in a specific service (old-time radio) for a specific type of part (air core inductors). Use it at your own risk.

So the one diameter away thing is for air core inductors, and if the shielding is really conductive it doesn't de-Q them as badly (although it'll still lower the inductance). And "good shielding" is probably something like 40dB down, because the old radio manuals where I see that recommendation also have all the wires laid out by hand, point to point between terminal blocks, and other practices that we would blanch at today. "Good shielding" means putting a piece of tin can between sensitive sections, without ever enclosing them.

Come to think of it, if you think of a helical resonator as a coil and a shield, then you have an example where shielding will _increase_ the Q dramatically, by reflecting energy that would otherwise be radiated back to the "inductor". Electromagnetics can be weird that way.

So, what are you using, just how much coil-coil coupling do you need to avoid, and do you need to shield the magnetic or the electrical fields?

You can make an electrostatic shield from a bunch of parallel wires, tied to ground at one end only, that can be pretty effective and only affects inductor Q proportional to the wire width (or maybe the square thereof). But it doesn't block the magnetic path at all.

Like Joerg said, placing inductors orthogonal to one another helps. That gives you three choices, if you have enough freedom of placement. For cored inductors you can probably get away with placing the shields much closer without bad effects. Inductors with closed cores are going to be self-shielding, so ungapped E-cores, toroids, cup cores, &c., are going to be better.

I suspect, but do not know, that even a gapped E-core is going have a smaller problem area than bobbin-wound inductor. I also suspect, but do not know, that a gapped cup core is going to have pretty good characteristics, if the gap is inside. I have always wanted to, but never have, experimented with making lightly coupled transformers by gluing toroid cores together edge to edge ('OO'), for old-time radio style IF transformers.

Use toroids. Watch out for saturation.

-- Boris

There are sometimes good reasons not to use toroids (drift, saturation, expense), and they don't shield perfectly.

But yea -- put toroids high on your list.