Switch mode DC-DC Converter EMI Input Filter capacitors

[transcendent]

I've been reading up about EMI filter design but have one key question I haven't got my head around yet.
A board I'm designing has a DC-DC converter brick, I would like to reduce the EMI radiated by long battery leads which feed its input.
The datasheet just specifies some bypass capacitors at the input and I would like to know:
How does the capacitance stated by the design affect my filter network parameters?

If I, for example, wish to put a parallel damped LC filter in front of the SMPS, does it matter that the input already has ceramic caps+large bulk electrolytics preceeding the filter?

This is probably a pretty stupid question but it is what it is..

 

[scopeprobe]

Any existing capacitance prior to the DC/DC is usually to provide a small current loop and low impedance source of power to aid stability which if left in place will form part of the filter characteristic. A properly designed filter can often replace these parts to optimise the impedance of the filter but if left in place and not be considered the attenuation and damping requirements may be different from that expected

 

[cupoftea]

Also, beware that if you suddenly plug in Vin, then their will be inrush into that existing capacitance you describe.........and now, you have placed your LC filter upstream of that capacitance......so that inrush will go through the inductor, and ring up with the capacitance, and possibly overvoltage the chips etc.

But otherwise, it sounds ok.........there is some rule called middlebrooks theorm, which describes how the smps shoudl be able to "look back" toward its input, and see a filter impedance which is lower than.......i forget the expression now, but its gotta be low impedance...so you need some decent capacitance just before the smps power stage.....as long as you have that then you can put what you like upstream of that, as long as you have that good bIt of capacitance just upstream of the smps input....which it sounds like you have.

 

[Warpspeed]

Low impedance suggests ferrite beads might be more effective and with fewer problems than a damped series inductance.

 

[KlausST]

Hi,

from the first post I read:
* SMPS, EMI, input capacitor, long cabling

these are raw informations but no values one can calculate with. So either you nned to give us some numbers or you need to do all the math on your own.

SMPS:
a good SMPS comes with a good datasheet containing EMI informations, switching frequency, voltage, current and not often even with EMI filter recomendations.

EMI:
What is yor most concern about EMI nooise in your application: Voltage related, common mode current, or differential mode current related

Long cabling:
The length is one parameter, but there are other (maybe even more important) paramteres. Especially the wiring method, cable type (shielde, twisted...)
Worst case surely are independent wires for plus and minus.
A much better way is to twist the single cables to a pair.
Or use a shielded cable.

Capacitor:
every capacitor type is good for a dedicated frequency range only. But even a well chosen capacitor is almost useless when wired the wrong way. Worst case is using single wires connected as a branch.
Much better is an "in line" connection, as close as possible at the SMPS with very close supply and return path.

***
Inductors:
Inductors can improve this situation. But they need to be deisgned as a whole system. .. including all the capactiors and damping circuitry.

***
All in all good EMI filtering need some understanding on HF, how antannas work, series and parallel impedances, wiring methods. But it also need an understanding of your unique situation.
Example: You talk about battery, thus I think about a car. A car is a bit of special situation because of its (usually) metal frame that is used a minus. Thus it´s a very different situation whether you talk about the "car inside" or the "car outside" EMI... and the frequency range ... and whether you use the frame as return path or use a wire/cable as return path.

Bad thing for you: There are so many situations and parameters ... thus there is no "one and only" good solution.
Good thing: The more you get experienced with EMI the less "trial and error" and the simpler get the solutions.

My recommendation: if you are not that experienced with EMI yet, then spend the time (and money) to get an expert for your situation. An expert that gives you the information what to care for - especially to learn for the future.
The more early during development you care about EMI the more simpler get the solutions. If you design the SMPS on your own .. the shorter ar the "antenna loops", the more effective, the cheaper, the simpler are filter solutions.

Sorry for the long text. EMI is a complex field. It is like driving a car. Stressful at the beginning. Not only throttle and brake. Many different situations like speed, road conditions, wheather, other drivers on the road, mistakes of the other drivers....But the more experienced you are the more relaxed you will drive. Keeping distance, stay concentrated ... and not over-react to problems (= synonymous to "damping resonances in filters") are good methods..

Klaus

 

[cupoftea]

Dont forget to address both common mode and diff mode emissions.
Eg use common mode chokes for Common mode