Differential versus common-mode SMPS EMC filter

[butterworth]

I notice in the book "power supply cookbook" by Marty Brown that he concludes every smps design example with the design of a second order common mode EMC filter.

He states that if this isnt sufficient a differential filter stage should be added to make it 3rd order.

I would have thought that the differential filter stage was the most crucial one in an SMPS and should have been the starting point.
-Since the differential filter is the one that actually filters the power switching current -those strong current harmonics which will otherwise give bad conducted emissisons.

The common mode choke will not filter these harmonics. It only filters that bit of energy which gets capacitively or inductively coupled to earth from the SMPS PCB.

So i am wondering why the common-mode filter is given presedence in this book?

(It seems worth remembering that in mainland Europe, the mains is only two-wire, hot & neutral, with no earth....and a SMPS connected to this would have no earth connection , and a common-mode choke would be useless.)

 

[Comcokid]

I would say it depends upon the application, market (i.e. regulatory agencies involved) and other factors.

A differential mode filter provides a degree of common mode rejection. I have examined designs where a lot of differential mode was incorporated, and no common mode was present.

I'm presently designing a push-pull converter for a specialty application. Yet, when I've looked at some of the 'overview' power supply books, the entire coverage of push-pull might be one paragraph indicating it's a special case of a buck topology, and continued with flyback converters as if the entire world consisted of flyback. I wish such authors would add to the subtitle indicating that the primary topic is flyback, so I would know to not buy the book.

Some authors become focused on doing things their way. Good authors will provide alternate approaches to a design.

 

[butterworth]

Good luck with the book situation, come to think of it, i havn't seen too much on push-pulls either.

I agree that differential mode filters provides some common mode rejection.

This is interesting though, on page 5 of this application note (half way down on the left)....it says that differential mode filtering is implemented as a common mode filter........

http://www.onsemi.com/pub_link/Collateral/AND8182-D.PDF

....that one really has me confused.

 

[zappedagain]

An antenna transmits a common mode signal, so unless you want to radiate on your cables and fail emissions testing, you need the common mode filter.

 

[sreid]

The problem with the Push-Pull type switcher was what was known as Flux Walking. At the time, the two NPN power transistor could end up with different On-Off times creating a DC bias on the transformer that would drive the core to saturation making then a big bang. FETs would probably have much less of this problem.

 

[butterworth]

i believe you can mitigate flux walking with a series cap. Or use current mode control. I think in days gone by comparators weren't quick enough to do current mode control.

Flux walking can also be a problem with half and full bridges. I tested a push pull once and one fet had been connected SDG instead of GDS.....it still worked and passed all the tests. We reckon one fet was repeatedly "flashing over" but it worked allright. it was only a 10w output. The FETS were also big well de-rated fets for some reason that i never found out