Martin wrote:

I am wondering if anybody has any experience with using Ferrite beads and EMI filters in a design? Items such as Panasonic EXC-CET103U(EMI) and EXC-CL4532U1(Ferrite).

Hallo Martin,

go to the website of 'Murata' at

http://www.murata.com/

There, they have beautiful application notes and technical articles, which in detail show how a design benefits from such filters.

You can also have a look at my reply

http://www.8052.com/forum/read.phtml?id=68096

where ferrite filters are shown in action.

When I'm using ferrites in my designs, then I often take ferrite bead 'BL02RN2' from 'Murata' or simple 6hole-ferrite choke '74275043' from 'Würth Elektronik' for power supply filtering of digital and/or analog chips. Also directly at input of 5V regulator I use such a part.

But don't worry about this filtering: Extense use of ferrite beads is only needed in special cases. Mostly the use of decoupling capacitors is enough.
Ferrite beads are needed, if you have to power very sensible circuitry by a heavily contaminated power supply voltage, like in the above link. This is often the case, when using an ADC in combination with digital circuitry (microcontroller, etc.)

If the micro is operating at 2 MHZ does that mean the frequency curves of the Ferrite or EMI filter should be selected to have largested attentuation close to 2 MHZ range?

No. Don't forget the harmonics! The higher the frequency of harmonic, the easier it is to produce unwanted radiation and the better the dampening of filter should be. Also, if a harmonic of supply current must run over a copper trace then the according voltage drop across parasitic inductance is the higher, the higher the frequency is. So, a ferrite bead should have high impedance up to 1GHz. Some types, showing a maximum in the 1...10MHz range, combined with a heavy fall of impedance at higher frequencies are not suited!
On the other hand, a ferrite bead should also show a high impedance at frequencies arround 1MHz. But then rather big and volumed parts are necessary. So, always a compromise must be choosen.

Ferrite bead alone is no filter. You also need a capacitor and some resistance in series with ferrite bead, unless you can allow some ringing. If the following relation is valid

R > SQRT (2 x L / C)

no ringing will occur. In order to keep R small, because DC supply current will cause unwanted voltage drop across it, from the relation results, that C should be choosen in the microfarad range.

How to determine L of ferrite bead?
If we take C in the microfarad range, resonance frequency will be at frequencies much lower than 1MHz. L in this range can be roughly estimated, when having a look at impedance curve: As long as ohmic losses (R) are negligible, impedance curve follows Z = w x L relation. If we read the impedance at 1MHz (Z1), for instance, we can estimate L by the help of

L ~ Z1 / 2 / pi / 1MHz

This estimation shows, that L of 'BL02RN2' from 'Murata' is about 2.2µH at low frequencies. 6hole-ferrite choke '74275043' from 'Würth Elektronik has an inductivity of about 20µH.
So, if we take C=10µF, then in combination with 'BL02RN2' R > 0.7Ohm should be valid. This resistance is so small, that even without added R no ringing will occur. Why? Because the equivalent series resistance of electrolytic capacitor of about 0.5...1Ohm will provide the needed ohmic losses.
When combining 10µF with 6hole-ferrite choke, R > 2Ohm must be valid. Then, without added series resistance some ringing at about f = 1 / 2 / pi / SQRT(L x C) = 11kHz must be taken into account.

In most cases, I only use a ferrite bead directly at input of 5V regulator, to prevent spreading of digital noise over the supply voltage. Of course, this filtering also prevents noise from outer world penetrating microcontroller application.

Up to now we only discussed the use of ferrite beads for power supply voltage filtering. But also filtering of digital signal lines (data buses, etc.) is possible. But again, the use of ferrite beads is only necessary in the most demanding cases. Especially, if digital lines are leaving the printed circuited board.
Then, in order to prevent heavy radiation, decreasing of slew rate to the allowed minimum is helpful. It's obvious, that then C of according LRC low pass filter must be kept rather small, may be smaller than 100pF, or so. But then, only rather small ferrite beads with rather low L can be used, otherwise heavy ringing would occur.
Assume 'BL02RN2' is used in combination with 100pF. The R > 210Ohm is needed. But with 6hole-choke, R >630Ohm would be needed to prevent ringing.
In many cases digital line filters are accomplished by simple RC filters, using up to some hundreds of Ohms in combintaion with up to 100pF. Sometimes, only resistances of about 10Ohm...100Ohm are used. But then, it's more the methode of matching source impedance to characteristic impedance of digital signal line (transmissionline), than to introduce low pass filtering.

Even if the use of ferrite beads might promise lots of benefit, never forget, that such kind of filtering meakes only sense in combination with solid ground planes!!! If, for instance, the ground return current runs over a copper trace of several hundreds of nH, then ferrite bead looses it's power, because of voltage dividing caused by both inductances. Only if the inductunce of ground return current is heavily smaller than the inductance of ferrite bead, filtering makes sense. Such low inductances of ground return currents can only be accomplished by the help of solid ground planes!

What about connecting analog and digital ground together at one point through a Ferrite.

Well, there are some 'philosophies' arround. But, if you take into account, that even a separated analog ground plane is never free of digital noise, inductance of this ferrite would cause a floating of analog ground relative to digital ground at high frequencies and would result in lot of common mode noise. Your ferrite would tend the concept of 'central star point grounding' ad absurdum.
Nevertheless, it's important to keep the connection point of both grounds free of noisy (digital) currents. And using ferrites with the power supply filtering, besides other measures, like shielding and grounding, will greatly help to provide this.

Bye,
Kai