question on EMI filtering

Thread Starter

weide83

Joined Jan 28, 2008
8
Hi all,

i currently working on power regulatoring circuit for a transceiver, i am trying to reduce the EMI of my switching power regulator. i was told that the EMI can be reduce using a LC filter and by carefully calculating the capacitor and inductor values. is it true?

Do anyone has any material on this??

Thanks
 

beenthere

Joined Apr 20, 2004
15,819
The output already has an LC filter. Adding more inductance may have an adverse effect if large currents are expected. Some significant reduction in the high frequency ripple can be made with the addition of several low ESR capacitors and at least one monolithinc ceramic to bypass the ripple. The creamic would be in the range of .01 - .1 uF. The others would probably be tantalum with values of at least 1 uF. hard to guess without knowing what current is involved. Each one makes a low-impedance path to ground for the ripple frequency, so a couple aid in its elimination.
 

nomurphy

Joined Aug 8, 2005
567
As beenthere has said, it's very dependent on the type of circuitry and output current you are designing. Typically, ripple current should be 20-40% of the output current, but sizing the inductor per switching frequency and output current is critical to EMI performance. Also, try using a magentically shielded inductor (see TDK VLF series or Cooper/Bussmann DR125 or 127 series).
 

scubasteve_911

Joined Dec 27, 2007
1,203
I'm confused about this question. Isn't EMI totally different from ripple voltage?

EMI is electromagnetic interference and it deals with actual electromagnetic radiation produced and its ability to reject it. Good layout practices and inductor shielding are the most effective ways to reduce EMI.

For voltage ripple, I would look at your switching time (ts goes up, ripple goes down), input capacitor (projecting input ripple to output ripple), output capacitor (ESR and value of capacitance are important, ESR as low as possible). Do not increase the output capacitor to an obscene level due to two reasons: big caps have higher ESR, and bigger caps and inductors in the circuit means poor response time to transient conditions.

Steve
 

DC_Kid

Joined Feb 25, 2008
1,072
muRata bnx022-01 or bnx002-01 (but they have tons of product to choose from)
this is what i use to filter my circuits in automotive applications. for me its just easier to buy the engineered product (if available) then trying to build it.
 

nomurphy

Joined Aug 8, 2005
567
I'm confused about this question. Isn't EMI totally different from ripple voltage?
Note I said ripple current, not voltage, the dv/dt of the LC circuit (the charging voltage vs. capacitor ESR) will create EMI. Optimizing the circuit per operating frequency and current will help prevent it, also adding an appropriately sized series ferrite on the output won't hurt (or a secondary LC filter). Using a mag shielded inductor can be critical to preventing both radiated and conducted emissions.

Along with signal intregrity issues where the signal is not trapezoidal and has a very fast edge or ringing, ground loops, and interrupted return paths where the ground is not contiguous with the signal (to name a few).
 

Thread Starter

weide83

Joined Jan 28, 2008
8
My transceiver will be drawing around 80mA(Rx mode) to 300mA (100% TX mode).

i still deciding which regulator to be use to power the transceiver. Hopefully a low noise one.
 

scubasteve_911

Joined Dec 27, 2007
1,203
Are you using IC transcievers? If so, you should check the PSRR vs. frequency.. I thought you decided on a switching regulator. I don't think a true low-noise switching regulator exists, well, in comparison to a linear device.

Steve
 

nomurphy

Joined Aug 8, 2005
567
It is still not a good idea to use the direct output of a switching PS to power sensitive analog circuits. Either use a post linear regulator for the analog section, or go linear in the first place. This may mean that the switcher generates +xVD (digital, or less sensitive) and also ++xV, that can be regulated down to +xVA (analog, or +xVQ for quiet).

where x = some voltage (5V, 12V, whatever).
 

Thread Starter

weide83

Joined Jan 28, 2008
8
i am trying to look for a efficicent regulator as this application is running on cells. therefore i trying to look into the possiblibity of switching regulator.
 

scubasteve_911

Joined Dec 27, 2007
1,203
It's hard to get the best of both world's.. This may be a situation where you need to know exactly what sort of noise is going to interfere with your transciever.. As I had mentioned earlier, the PSRR curves might give you some more insight. If you design a relatively low-frequency switcher and your PSRR vs frequency indicates that it can handle the fundamental switching frequency along with a few harmonics, it may not be necessary. On the contrary, if you find it does affect it too much, then you may need to design the layout very well or switch to a lower EMI switching typology.

Steve
 
Top