Choosing an inductor

[aaronhance]

Hello, I'm a Software Engineering student and an electronics hobbyist. I am currently working on designing a 3d printer controller board with a free-scale MCU that has a ARM cortex-m at 100Mhz. I'm looking at some reference designs and for high frequency noise reduction at the power input it has some inductors, but it fails to mention what values they should have. What values would be good for this situation and why? Thanks!

[tautech]

Welcome to the forum.

Mains filtering usually just consists of a common mode choke and class X and Y caps.
The choke values may be determined by their current carrying capability, this we would need to know.

[aaronhance]

Welcome to the forum.

Mains filtering usually just consists of a common mode choke and class X and Y caps.
The choke values may be determined by their current carrying capability, this we would need to know.

Hi, it's on the input pins for the mcu, not the power supply itself.

[pmbrunelle]

I would put ferrites there.

There's not too much capacitance, so if you put more proper inductors, you would risk having a high Q, and then the response to sudden load demands wouldn't be too good.

[T3sl4co1l]

Ferrite beads are generally unsuitable for power (due to subtle and not-widely-known reasons), but can still be okay if the current draw is small (under 20mA, give or take?).  (The current the datasheet tells you about is only the thermal limit, not electrical.  So it's kind of meaningless, unfortunately.)

Inductors, however, are sold by [electrical] current rating and value.  So if you have to filter more current, expect to shop for an inductor, not a FB.

Noise and filtering are also quite complex subjects, with the worst part being that radio frequencies tend not to stay in wires very well...

Noise is always a question of: is it too much?  How much does it produce?  How much does it need to be?

Perhaps worst of all is that, schematics represent only the components the circuit is built with.  The PCB layout is a complex RLC component all its own, and deserves much attention, especially for circuits with high frequency content (like radios, switching supplies, digital logic and so on).  The PCB has to be designed to "keep RF in the wires".

If all you're doing is digital, I wouldn't worry about it.  Place bypass caps near ICs, use ground pour wherever possible (stitched with vias), and if you have long, high speed connections (these days, pretty much any CMOS chip has pretty fast outputs -- whether you need each and every nanosecond, or not!), make sure they're adequately terminated (usually by placing a resistor in series with each output/driving pin, as close to the chip as possible).

Note that short risetimes equate with high frequencies, so fast digital signals have a greater likelihood of "not staying in the wires".

If you have some analog on the board, try to keep digital signals away from it.  Don't try to cram everything together, segregate by region and function.  Add filtering to the supply rails, where they enter/exit each region.

Finally, if your required noise level has a hard limit -- such as a product sold for the U.S. or EU market (which must meet FCC Part 15 or CISPR 22 limits, respectively), expect to take a lot of time at a testing lab (and consider hiring someone to do the busy work!), getting that noise within limits.

Tim