Selection Process for Common Mode Transformers for smps

[D from BC]

I've noticed that most smps have a common mode transformer in the line
filter section.

I looked at this spec sheet on:
http://www.tdk.co.jp/tefe02/e93_uf.pdf
as an example.

At the moment I'm clueless on selecting a CM xformer for my
application and I'm tempted to fudge the selection..
Can I go for the maximum inductance until it takes up too much space?
Maybe a rule of thumb exists...say 1/4 the size of the power magnetics
in the converter?
I know this is crappy..I probably should know my noise level, target a
CSA conducted EMI limit, get my attenuation and then pick a CM
xformer.
With a working sim of my smps, can I estimate the required CM xformer
specs in LTSpice?
Whats the general design procedure for selecting a CM xformer?

D from BC

 

[legg]

I've noticed that most smps have a common mode transformer in the line
filter section.

I looked at this spec sheet on:
http://www.tdk.co.jp/tefe02/e93_uf.pdf
as an example.

At the moment I'm clueless on selecting a CM xformer for my
application and I'm tempted to fudge the selection..
Can I go for the maximum inductance until it takes up too much space?
Maybe a rule of thumb exists...say 1/4 the size of the power magnetics
in the converter?
I know this is crappy..I probably should know my noise level, target a
CSA conducted EMI limit, get my attenuation and then pick a CM
xformer.
With a working sim of my smps, can I estimate the required CM xformer
specs in LTSpice?
Whats the general design procedure for selecting a CM xformer?

If you know the amplitude of the lowest relevant harmonic currents
being pumped into the environment, and the maximum common mode
capacitance permitted (safety leakage current implications) , you can
establish a minimum L value required to meet intended interference
levels (by an acceptable margin) in a 50 ohm LISN receiver.

This L value will be more difficult to achieve as the power
requirement of the load (and the line current) increases. Heavier
gauge wires take up more room, larger cores are more expensive and
higher frequency performance will be more easily compromised. At
higher peak line currents, the high permeability core may saturate on
it's leakage flux. At higher temperatures the core saturates at lower
leakage flux levels.

At some stage it may become more economical and effective to use two
inductors in a two-stage filter.

Physical iterations involving realistic physical representation of the
final product, at the earliest stages possible, are highly
recommended. Models don't necessarily suck, they are just
first-order approximations resulting from an inevitably inaccurate
schematic.

RL

 

[D from BC]

Thanks...
I do some googling on LISN and max CM C..
Also I'll think about the all magnetics parameters when selecting a CM
xformer.
D from BC

 

[Terry Given]

Thanks...
I do some googling on LISN and max CM C..
Also I'll think about the all magnetics parameters when selecting a CM
xformer.
D from BC

you can also get some CM cores that are designed to have high DM
inductance as well - I used one to fix a customers design that failed EMI.

If you dont know how much CM L you need, fit the biggest one you can,
and hope for the best

Ideally you calculate the CM and DM noise at the design stage, and
select an appropriate part (as Legg said), then (using a DMRN or
similar) measure the CM noise at the prototype stage, and adjust to
suit. If you include SRF in your calculations the results can be quite
reasonable - until, of course, layout parasitics rear their ugly heads!

I have a couple of MeanWell SMPS here that have two DM/CM chokes. And a
layout designed to maximise EMI....

Cheers
Terry

 

[D from BC]

you can also get some CM cores that are designed to have high DM
inductance as well - I used one to fix a customers design that failed EMI.

If you dont know how much CM L you need, fit the biggest one you can,
and hope for the best

Ideally you calculate the CM and DM noise at the design stage, and
select an appropriate part (as Legg said), then (using a DMRN or
similar) measure the CM noise at the prototype stage, and adjust to
suit. If you include SRF in your calculations the results can be quite
reasonable - until, of course, layout parasitics rear their ugly heads!

I have a couple of MeanWell SMPS here that have two DM/CM chokes. And a
layout designed to maximise EMI....

Cheers
Terry

Yes!! If I stay under the OEM CM xformer Irating and then I can go
for max L until I can't tolerate the physical size..

I was fishing for 2 solutions...
The do it fast solution: Use biggest L as real estate allows, cost not
important, gamble on noise rejection.
The do it right solution: Calculate, measure, select a part and test.

Thanks.. I'll look out for the CM/DM xformers as well..

D from BC