EMI Suppression / Common Mode Choke Questions

Hello SWFlys,

This is really an excellent question.

• Is it better to place the common mode choke on the un-rectified AC voltage, on the rectified DC voltage, or both?

To begin with, a CLC pi filter has more attenuation on filtering nosie than the common LC filter. you can easily find materail about the difference between LC filter and pi filter. So normally, if you need maximum attenuation on EMI and you have two input cap, you can put the common mode choke in bteween to act as pi filter like DER531. As for AN22, we only have one input cap, so we put the common mode choke at the AC side, of course you can use two cap at input and put the common mode choke in between. depends on how much filter you need, you may need both in some design to put the maximum common mode filter.

• Does the optmal EMI filter design ultimately come down to trial and error? (I hope not...)

If you had the accurate model of your circuit, of course you can calculate the optmal value. but for the time and resourse concern, you try and see the performance will be more efficient.  

• The critical characteristic of the common mode choke is not strictly its inductance, but its cutoff frequency, determined by its inductance and series resonance (R / 2pi L), correct? (And, of course, its saturation current)

Partially right. the resonance also depends on the indutance L. 

Best Regards

It turned out that I have no 18uF so I put 2 * 22uF400V to input filter stage. Have I change the CMC's inductance somehow in order to maintain proper filtering ability of the input filter (I mean practice not theory)? Can I use 2 * 22mH*1,5Ohm readymade CMC without trimming its inductance? Will bigger inductance work 'better' or 'worse' than nominal one? Thank you!

the Common Mode Choke (ONE of its two bobbins separated by partition of the plastic frame). Am I right?

Is it useful / permissible? Can I trim the CMC's inductance by means of gap thickness?

Hello Alex,

It all depends on how much filter you need. i don't know how much EMI performance you have and what's your requirment. The best or maybe easest way for you is to measure the EMI performance of your board to see if it meets the requirement. if not, you should try to use bigger CMC, otherwise it is no need to use the big CMC for the size and price consideration

Best Regards

Reference design RDR-469 includes the capacitor C9. I'm right that C9 and L1 are part of the EMI filter for CM noise? Why is the primary side of the capasitor placed on the "hot" side of the transformer T1 and not on the source of the INN2605K?

Hello Alex,

yes you can if your inductor allows you to do so.

Best Regards

Hello SW-EVS,

You are right.  C9 and L1 are part of the EMI filter for CM noise. You can also put it on the source of INN2605K for the CM noise consideration. 

Best Regards

According to your answer, please can you explain me the comment in the data sheet:

Y Capacitor
The placement of the Y capacitor should be directly from the primary
input filter capacitor positive terminal to the output positive or return
terminal of the transformer secondary. Such a placement will route
high magnitude common mode surge currents away from the
InnoSwitch-EP IC. Note – if an input π (C, L, C) EMI filter is used
then the inductor in the filter should be placed between the negative
terminals of the input filter capacitors.

Thank you very much!

Hello SW-EVS,

What i mean is for the EMI Common mode noise consideration, you can put the Y cap at primary source side. 

In our datasheet, we suggest the Y cap directly from the primary input filter capacitor positive terminal to the output positive or return terminal of the transformer secondary. That's for Common mode combination or ringwave surge consideration. We want to keep the source of the IC quiet and don't want the heavy current during surge go through the source trace. we suggest the Ycap at the positive side. 

The pi filter inductor is high impedance for the common mode surge current. For the same resaon, we suggest to put it at the negtive side to block the heavy current into the source trace during common mode surge.

Best regards