0
\$\begingroup\$

How should inductors be placed on a PCB? I have a relay (MCHMR1-S DC24V) switching about 3 A of current. Do I need to separate the relay from nearby analog and digital components, or is it fine next to either?

Additionally, can shielded inductors be placed next to each other (or near any other analog/digital components)? These inductors are not used for power switching, they are used in an audio application (not the relay). Both are 47 μH, 2.5 A, through-hole inductors.

\$\endgroup\$
1
  • \$\begingroup\$ what in audio needs a 47uH 3A inductor, if it's not in the power amp? If they carry signal current, they can in principle couple thr signal to other nearby inductors and even to low impedance traces. But you would need to focus your question more for a precise answer. \$\endgroup\$
    – tobalt
    Feb 18 at 6:12

1 Answer 1

2
\$\begingroup\$

Distance affects the coupling coefficient between inductors.

Most generally, one can consider all inductors coupled, that coupling given by a matrix of coupling factors, from each one to every other. Then, understand most of the matrix entries are very close to zero: close enough to approximate as actual zero for engineering purposes (e.g., interference is below noise floor for the audio signals, or below the emissions limits for EMC purposes), and thus we can work with it as a sparse matrix, or further subdivide it into groups of relatively-strongly coupled inductors, which may include groups of size 2 (mere adjacent pairs). We might also approximate small groups as pairs, or chains of pairs, to simplify analysis, when it is appropriate and convenient to do so.

A full matrix approach is severe overkill for most purposes, but can find use at high frequencies, where indeed the whole PCB, or PCBA + wiring harnesses + enclosures, might need to be treated as a coupled system. Matrix extraction software is available (e.g. ANSYS tools).

Mostly, we take the converse approach, designing the system from the bottom up, in such a way that the matrix will be near-sparse at frequencies of interest. In most commercial designs, this is done with a ground plane PCB strategy, or a metallic enclosure if needed; in RF designs, this is done by adding shield cans, or indeed solid metal covers (with pockets milled out to clear components and filter structures), to physically block fields between sections.

Notice a sensitive RF transceiver will have a much tighter interference limit, and coupling factors are more significant at high frequencies, therefore more aggressive countermeasures are chosen. The interference threshold is a key part of the design process, and dictates what shielding strategy is chosen. The threshold is different for different parts, signals and applications.

On a PCB, typically the interference thresholds and coupling factors are such that we only need to be concerned about those that are within a couple bounding spheres of each other, that is, taking the maximum outer dimensions or bounding sphere around each part, expand it by a factor of 2 or 3, and considering all within that space. And we can thin that down even further depending on part type, signal, etc.

And, plenty of situations hardly care about coupling at all; most power applications won't care about coupling between inductors if they're heavily-loaded filter inductors (rod core type chokes are a common choice for DC output filtering), or doing similar things (the inductors in a phase-interleave converter might indeed benefit from coupling, there are a few papers on this topic). Or that the interference limit is high enough it can never be met through ordinary component placement (e.g. a wound-rod style ferrite bead on a 5V logic signal (~1.5V noise margin) beside an unshielded filter inductor on a power rail having <1V of noise/ripple).

Shielded type inductors are generally lower in external fields, though it's not a controlled commercial term and many parts are dubiously or loosely labeled as such. Conversely, due to geometry and relative orientation, there are unshielded types with lower external field, or less impact upon nearby parts of different orientation, compared to shielded types. It's a "nice to have" term, but not very meaningful for design purposes.

Also, inductors are never rated in terms of fields or emissions, so you'll have to test samples to know absolutely for sure whether they will work in a design of given interference limit.

Finally, regarding relays: the coil is generally well-shielded, on account of its magnetic path being constrained by the pole pieces and armature. What's more, the voltage is constrained by the flyback clamping circuit used, and the turns ratio is extremely low to anything nearby (the coil might have thousands of turns, and a couple tens of volts on the coil is reduced to 10s mV/turn). A far greater concern is switching current, as mechanical contacts act extremely quickly (sub-ns) and closing and opening under load can both generate intense electromagnetic pulse (see standards such as IEC 61000-4-4, electrical fast transients). You may want a snubber for the contacts, to avoid interference in the rest of your circuit.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.