An FB is just an impedance -- check the datasheet (if it goes into any detail, that is). The impedance is largely real at higher frequencies, meaning it's resistive, lossy, self heating.
Note that FBs saturate at very modest currents (even large "6A" parts die off in the vicinity of 100mA), so the impedance you're inserting need not be as useful as the headline claims.
If you throw them in, willy nilly, it's very likely you're using them wrong. They're not usually enough inductance to matter in terms of low frequency response, and with the low saturation, that inductance is only manifest at small signals anyway. So a power inrush surge, for example, might saturate it to the point where it looks like a piece of wire of equivalent length (~10s nH?), and the power supply network comes up as if it were made of wire (little overshoot, well dampened by large bypass caps).
So, they might be applied "wrong", but that's usually harmless (which I suppose is good business for both FB manufacturers and appnote writers).
As for absorption, it can be beneficial. Suppose you light a room with a white bulb in the middle. Suppose you don't want any of that light to get out. If you construct the room entirely from polished aluminum panels, it will be very well lit indeed inside that room, but little light will escape. If you use matte black panels instead, it will be dimmer inside, and still, little light will escape. But if you have the possibility of radiation from that room -- the panels have holes, or don't quite overlap perfectly -- which are you better with, the shiny or the black version?
The exact same goes for filters, except one-dimensionally within wires (hopefully..) rather than in a 3D box. The random holes or gaps represent possible radiation-susceptibility paths in your circuit. If you're reflecting all that energy (with a high-Q filter), it stays bottled up on the SMPS side, which is a good thing. But that also potentially makes it "brighter", too -- more current bouncing around the switching components and filter capacitors. A purely dissipative filter would be very lossy indeed (you don't want to absorb all the reactive energy from the switcher, because it needs that to operate at high efficiency!), but some compromise is clearly desirable. This is very broadly where ferrite beads come in -- they're cheap as hell, handy enough, and do a reasonable job of breaking high frequency resonances.
Because ferrite beads saturate so easily, they're best confined to small signal traces (analog and digital inputs and outputs), and common mode ("current compensating") roles (beads around cables, common mode chokes).
Often, the resistance is used explicitly: suppose you have a main board attached to a display board with a ribbon cable. That cable has some equivalent inductance, and the boards have some equivalent capacitance; together, they resonate as a dipole structure, at some characteristic frequency (maybe in the lower 100s MHz). Over most of the band, the structure has a low impedance, and nothing couples into your ribbon cable's signals (say you're passing around some 5V CMOS logic signals, so you have a good 2 or 3V of headroom before logic thresholds start going completely nuts). But at that one characteristic frequency, give or take a few percent (it might have a Q factor of 30ish?), the impedance spikes, and the voltage between boards, and the current through the cable, becomes huge. Suddenly, you've got, I don't know, 30V between boards, from a modest 3V/m external stimulus, and some fraction of that 30V will not be shared evenly between all the wires in the cable. And it only takes 10% of that to blow your logic levels and get gibberish, which will easily be provided (the coupling factor between lines in a ribbon cable isn't terrific, even if you grounded every other signal -- which you should do for signal quality purposes anyway).
Now, add a ferrite bead to that cable, maybe a modest 100 ohms near that resonant frequency. Now there's a modest impedance over a wide range of frequencies, which means it will act like a shitty antenna over a wide range. The induced voltage will absolutely be larger at other frequencies -- but, still well within the logic threshold. (The induced *difference* won't actually be worse at all, because the ferrite bead acts in common to all signal and ground traces.) And at the peak, it will be severely dampened, so your worst case scenario goes from "oh shit" to "meh"!
So, FBs are best around cables, and not so much for filtering purposes, but for dampening purposes, which improves on that worst case scenario. FBs aren't great for filtering -- if you wanted a good filter, would you prefer an L || R over a high Q inductor of the same value? Hell no! But the attenuation can have knock-on value, like dampening parasitic modes and radiation (even within filters of much lower cutoff frequencies).
Tim
Home
Help
Search
Login
Register
