Passing EMC, filters and more . . .

We are commonly asked to assist in solving EMC filtering issues.

Solving conducted emissions to better than 3dB on your power supply or inverter is a great milestone to make on any project ( ideally class B ).

Not only because you now pass part of the EMC requirements - but it is likely that your filter will help you pass radiated standards too, as you have already reduced - somewhat - the noise that makes it to the phase/neutral, or output wires.

Amazingly ( to us ) most people seem to go about designing filters for EMC in a somewhat backward way.

First they "design" a filter board based on complete guess work, which they then fit to the power converter - then they measure the RF noise on phase and neutral and see where the bad points are - and then they try to modify the filter to suppress the high spots - and end up spending hours playing whack-a-mole with emissions spikes and generally getting nowhere at high speed.

This is not the most efficacious way to learn how to design a filter, nor to pass EMC.

Here is what we advise:

1, Design a filter based ( in part ) on the expected differential mode ( DM ) emissions from your converter, these will be DM chokes and caps - or you can rely on the DM component of the CM chokes you would like to use, and calculate the DM caps accordingly. DM emissions are closely tied to ESL and ESR of the main bus caps supplying your converter ( or the caps across the AC line / rectified line, for a booster/PFC stage ).

2, Find out what the expected common mode ( CM ) emissions will be, this can be tricky if your heat-sinks are earthed - or your earthed case is near to switching circuits when fitted. If you really don't know what this might be - get reading on the net, or ask an experienced engineer to assist. Integrate the desired amount of CM attenuation into your filter.

3, Now, it is a fact that every LC combination in your filter can act as a resonant circuit. Thus the DM inductance of a CM choke can resonate ( ring ) with a Y cap to earth, DM chokes will ring with DM caps ( and Y caps ) - so that you do end up with all sorts of resonances in your starting filter. [ until you measure a complex multi-stage filter - you really have no idea of its attenuation over frequency ].

4, So, now, using a spec ann with a tracking generator - we can measure the intended filter, both DM, phase to neutral, and CM, P+N to earth. It may be an idea to use a 1k + 1k divider from the driving source ( 50 ohm ) and take of factor of 0.5 into account. Measure at a connected LISN if you can.

We can do the same with a signal generator that goes from 100kHz to 30MHz and a scope ( 5mV ) - we just can't see results as easily as with a tracking spec ann ( nor at the low signal levels that would typically be present ). Remember to load the scope measuring point to at least 1k.

So we apply signal to the inside of the filter and see what comes out the phase and neutral terminals for DM ( P+N to earth for CM ).

Naturally we hope to see the attenuation of the filter as the freq goes up, but what we will really see is the inherent resonances of the filter causing peaking of the input signal at the output ( not helpful ), and places where the attenuation is higher than expected due to parallel resonance of both DM & CM chokes ( very helpful for repeatable choke manufacture ).

Remember that at frequencies above ~10MHz, inductors look like small value caps, caps look like low value inductors, and pcb traces are slightly lossy inductors - so don't be surprised to see odd things in the 5 - 30MHz part of the sweep.

Now we set about damping the observed resonances ( signal peaking ) that we have just measured. This will involve damping type ferrite beads on the Y cap leads ( or use low value R's ), resistors across the DM & possibly also CM chokes, and for extreme cases you may need to put RC dampers across some of the DM caps.

5, Once we have fixed the nasty resonances - YOU CAN THEN, AND ONLY THEN, FIT THE FILTER TO THE CONVERTER.

We now have some small measure of peace of mind - as we know that whatever RF noise comes out of the converter, our filter will likely not make it worse.

Of course DM chokes are not perfect, they have a self resonant frequency, SRF, above this frequency they are capacitors and let all the VHF noise straight through. Ditto for CM chokes such that we always get poorer filtering performance than we hoped for from a filter.

6, Now is the time to modify the filter to adapt the parallel resonances of the filter chokes to block out the higher peaks of the RF noise produced by the converter. Usually we find we need at least TWO CM stages, one for lower frequencies, one for higher. We may need to add special DM chokes in the phase and neutral lines to combat higher CM frequencies that a standard CM choke cannot touch.

For every modification - we need to re-measure the filter card as per 4 above - so that we do not inadvertently add a resonance that peaks up the noise.

7, It makes sense to filter the higher frequencies physically nearer the point of cable entry/exit, and to ruggedly shield any cables at the quiet end of the filter.

Now - there are a myriad of tips and tricks gleaned by our EMC engineers from over 30+ years of bench testing and field experience relating to construction of CM and DM chokes for noise rejection, how to use caps for very high frequency reduction and to reduce their pickup, filter board layout and placement, and much, much more.

For more of the good stuff concerning passing EMC, go to pwrtrnx.com