EMI Filter Design

[Nikos A.]

Hi everyone,

Is there any tool/guide/simulator to proper calculate the EMI filters? More specifically I want to design the EMI filter for a DC 12V converter and I am facing difficulties..

Thanks is advance!!

[tszaboo]

Würth has a tool for it's part
https://www.we-online.com/redexpert/#/home/article/emc-design-tools
And this is worth looking at, to figure out what you might need in the first place, depending on the frequency.
https://www.we-online.com/web/en/index.php/show/media/07_electronic_components/toolbox_1/Poster_Filter-Storage-Applikation.pdf
In the end it can be very difficult, or very easy, depending on what you want to do.

[T3sl4co1l]

Not really. So much depends on the noise source, what the load is connected to, the layout, the enclosure -- that a general solution will be too general to be useful.

In short, guess a reasonable starting value, and allow room for testing of different parts and values.  Try all permutations of Y capacitors.

With a great amount of experience, you can anticipate which configurations are likely to be the most effective, but it's still guesswork.

There are enterprise-grade simulators that can include circuit, layout and enclosure and cabling in a full EM simulation, and produce representative EMI results without building a prototype; but the software suite to do this costs way more than a few protos and a few weeks of lab time, for most users.

Tim

[tszaboo]

Not really. So much depends on the noise source, what the load is connected to, the layout, the enclosure -- that a general solution will be too general to be useful.

In short, guess a reasonable starting value, and allow room for testing of different parts and values.  Try all permutations of Y capacitors.

With a great amount of experience, you can anticipate which configurations are likely to be the most effective, but it's still guesswork.

There are enterprise-grade simulators that can include circuit, layout and enclosure and cabling in a full EM simulation, and produce representative EMI results without building a prototype; but the software suite to do this costs way more than a few protos and a few weeks of lab time, for most users.

Tim

It could be a simple DC-DC converter for all we know...

[T3sl4co1l]

What's "simple" about it?

There are perhaps fewer variations of "DC-DC converter" than of other kinds, but all of them have to fit inside that generalization; it's still a broad net to cast.

Tim

[Nikos A.]

Thank you for your answers!!

Well this is the  dc converter that I am intending to use:

https://eu.mouser.com/datasheet/2/687/tel5-519367.pdf

and unfortunately the manufacturer does not provide any filtering circuit..

I do not have any experience with EMI protection and I do not know how to start..

[awallin]

murata has good application notes, e.g.
https://www.murata.com/en-global/products/emc/emifil/knowhow/basic

as for components, murata BNX025 has been working OK for me after AC/DC brick-supply, before LDO.

[T3sl4co1l]

The datasheet claims it does not need a filter, or contains it internally, to meet CISPR 22.  As usual, it does not document how it was tested (example, with a load resistor and no other secondary side connection, you can get away with a lot; if it's common ground and both inputs are tested through a LISN, there is much less you can get away with).

Do you need to meet a more stringent level?  Do you know yet whether you have any EMI problems at all?

Tim

[pinkman]

This really depends heavily on your application and what standard you must achieve.  There are varying levels of requirement even within a standard, take CISPR 25 for example.  Will you be required to test your design and achieve a specific EMC/EMI certification? If so, which one?

Some have commented above that designing EMI filters this can be perceived as "guesswork", that is partially true simply because our minds cannot comprehend what is going on beyond the time domain and empirical data vs your modifications to the circuit may not make sense to you without simulation.  If you are not required to test, and have no access to some standardized test setup, you can certainly get an idea of radiated emissions with a relatively simple near field probe setup; you have probably seen these sets of 4 near-field probes that come with a small wideband amplifier.  This could be used with a spectrum analyzer to get a semi-reasonable idea of relative emission over frequency but expecting absolute accuracy would be unreasonable.  This is more useful for A-B comparison when making modifications.  For measurement of conducted emissions you would want a wide bandwidth current probe to start with and those can get a bit pricey.

To start with, placing PCB lands for a series ferrite of appropriate size for the output current (consider derating!) on the input and output along with associated capacitors would be a good start; If you end up with conducted emissions problems you may want a spot for a common mode choke on the input.

[Nikos A.]

Thank you for your answers!!

The datasheet claims it does not need a filter, or contains it internally, to meet CISPR 22.  As usual, it does not document how it was tested (example, with a load resistor and no other secondary side connection, you can get away with a lot; if it's common ground and both inputs are tested through a LISN, there is much less you can get away with).

Do you need to meet a more stringent level?  Do you know yet whether you have any EMI problems at all?

Tim

You are right, this specific converter is internally complied with CISPR 22 class A. What I've read is that class A is designated for industrial applications while class B for residential applications. This project is not intended to be used for residential application and thus I won't proceed with any further filter designing..

However, I would like to know what the limits are for each class and after searching I found this article by digikey that cites those limits.

https://www.digikey.com/en/articles/techzone/2012/jun/understanding-electromagnetic-compatibility-standards-for-switch-mode-power-supplies



Well, in case that I need in the future to comply the product with class B requirements it should be designed a filter to prevent conducted emissions above class B limits isn't?

Why the frequency of emissions in the table has a limit of 30MHz?