EMC and harmonic mitigation

[R_G_B_]

When mitigating harmonics using a lowpass filter .
And there's self resonance at a frequency much higher than your products operating frequency.

What part does self resonance play in EMC testing if the filter has done its job and removes  the intended harmonics
So that they fall with in the specified limits. But further on say 10Mhz the resulting self resonance peak of the filter means that any out side interference is not filtered at this self resonant frequency?

Does this make sense I'm.confused about the 150khz 30Mhz 80% modulation frequency sweep. Is this to test the effectiveness of the filter including the self resonance.

In other words the filter self resonance should be well outside the band of interest.

Sorry if this is a convoluted post

[penfold]

With the resonant frequencies of input filters, yeah, most filters have them, but perhaps the remainder of the DUT is fairly tolerant to noise at frequencies at this range, perhaps there's filtering going on elsewhere in the system which rejects it better. Good design in the rest of the system is one way to get around it, as it cascading filters to improve rejection at either filters' resonances.

I wouldn't necessarily word it around testing the effectiveness of the input filter, it's more about testing the response of the DUT. If you consider that the DUT may receive and "detect" the incident signal/wave, perhaps on an analogue input with some nonlinearity, the modulation can help reveal its presence since the "error" would appear as a 1kHz sinewave. I believe the modulation is supposed to make the test signal more representative of an AM modulated radio signal.

The modulation does present three frequencies (in comparison with an unmodulated carrier containing only one frequency), at f_carrier, f_carrier+f_modulation and f_carrier-f_modulation, but with a separation of 1kHz, there aren't many properties of an input filter that would be particularly revealed in that case compared with a non-modulated carrier. It can be a more intensive test of the DUT itself, as a system that would be able to reject a DC offset caused by "detecting" a continuous-wave incident field/signal, now has to contend with the 1kHz it will be demodulated as... so its not really anything specific to do with filter resonances, but indirectly, yes, they would let through more of the incident signal making an existing problem worse.

[R_G_B_]

Thanks for you post.

So I sort of  see what your getting at. And as you rightly say most filters have self resonance of some sort due to the components having some form of  capacitance. So the self resonance needs to be taken into account irrespective even if the filter attenuates  the harmonics to the set limits.

[jonpaul]

Very confusing question: Most EMI filters are designed to work with typical mains line equiv impedances and PSU inputs.

These low Z source and loads result in very low system Q.

Jon

[penfold]

So I sort of  see what your getting at. And as you rightly say most filters have self resonance of some sort due to the components having some form of  capacitance. So the self resonance needs to be taken into account irrespective even if the filter attenuates  the harmonics to the set limits.

I'd probably go further to say all filters of that nature particularly those designed to operate over a wide range of frequencies and operating currents etc will have some very noticeable deviations from ideality, may be resonances or plateaus, as you say mostly from stray capacitances between inductor windings or equivalent series inductances in capacitors etc.

You can't really (read as "you can but it's hard work") view EMC filter responses as "ideal plus parasitic", they are kinda just what they are... messy and complex responses. Finding one which works well for your application is more about understanding your system's particular susceptibilities and finding one which has resonances and big deviations that don't cause too much upset to your system.