The counterintuitive result
A ferrite bead is often reached for as a universal emissions fix: add one in series with the input, watch the quasi-peak scan drop, ship. But a bead is not a damping element in any automatic sense. It is an impedance that changes with frequency, material, and operating conditions, and when it sits inside an undamped LC filter it can move the filter’s resonant peak rather than remove it. The result is a scan that looks worse at one frequency while looking better at another, and a team that concludes the bead was “bad” when the actual problem was the filter topology around it.
This article is about that failure mode at 1k–10k unit volumes, where the prototype passed, the production build did not, and the difference is often a component decision made on a contract manufacturer’s floor rather than in the schematic. The argument is narrow: before you blame the bead, check whether the filter it lives in is peaking, and check whether the bead’s impedance at the actual operating point still resembles the datasheet curve.
What an undamped LC filter actually does
A second-order LC filter is not a gentle slope. Texas Instruments’ application note SNVA538 on input filter design states the behavior plainly: an undamped second-order filter “provides 12dB per octave of attenuation after the cutoff frequency f0, it has no gain before f0, and presents a peaking at the resonant frequency f0” (TI SNVA538). The same note warns that “the gain near the cutoff frequency could be very large, and amplify the noise at that frequency.”
That is the mechanism behind the counterintuitive result. If your switching converter has a noise component near the filter’s corner frequency, an undamped filter can make that component louder at the input terminals, not quieter. Adding a bead changes the impedance profile, but it does not necessarily add damping. It can shift the corner, add a second resonance, or interact with the decoupling capacitor to create a new peak.
The damping factor ζ (zeta) in the standard second-order transfer function controls how much peaking appears at f0. TI’s note shows that as ζ becomes smaller, the gain at the corner frequency becomes larger, and that a damping factor of 1/√2 gives 3 dB attenuation at the corner with favorable control-loop behavior. The note also flags a second consequence: “A poor damping factor on the input filter design could have other side effects on the final performance of the system. It can influence the transfer function of the feedback control loop, and cause some oscillations at the output of the power supply.”
So the bead is not the only actor. The inductor, the capacitor, the ESR (equivalent series resistance) of that capacitor, the converter’s input impedance, and the load all set where the peak lands and how tall it is.
Why a ferrite bead is not automatically a damper
A ferrite bead’s impedance is usually plotted as magnitude versus frequency, with the resistive component rising through a band and the reactive component dominating elsewhere. Murata’s guidance on selecting ferrite beads notes that “the frequency range where the impedance curve rises varies depending on the material of the ferrite bead” and that “the ratio of the resistance component R to the reactance component X of the impedance also varies” (Murata, How to Select Ferrite Beads).
That matters because a bead only behaves like a resistor—the thing that actually damps a resonance—in the frequency band where its resistive component dominates. Murata also notes that a bead type with a sharp-rising impedance curve “tends to have a narrow frequency band where the resistance component is the main component,” so outside that band the bead is more reactive than resistive. A reactive element in an LC network shifts resonance; it does not absorb it.
Two further mechanisms are widely discussed in ferrite bead application literature but were not supported by the sources retrieved for this article: DC-bias derating of bead impedance, and self-resonance of the bead itself. Both are real concerns worth verifying against the specific part’s datasheet and any vendor application note that plots impedance under bias. The point here is procedural: do not assume the datasheet curve at 0 A applies at your operating current, and do not assume the bead is resistive at the frequency where your filter peaks. Measure or obtain the biased curve before treating the bead as a fix.
The impedance-separation check
TI’s note invokes Middlebrook’s extra element theorem to explain when an input filter destabilizes a converter: “to avoid oscillations it is important to keep the peak output impedance of the filter below the input impedance of the converter.” The note’s Figure 3 plots the filter output impedance against the converter input impedance and shows the two curves should be well separated.
This is the check that matters more than “did we add a bead.” If the filter’s peak output impedance approaches or exceeds the converter’s input impedance at the same frequency, the filter is not just failing to attenuate—it is interacting with the converter’s control loop. The emissions scan may show a bump at that frequency, and the output voltage may show low-frequency oscillation that nobody attributed to the input filter.
The practical procedure is to plot both curves across the relevant frequency range: the filter’s output impedance (including the bead, the inductor, the capacitor, and their parasitics) and the converter’s input impedance. Where they cross or come close, that is where to look for the emissions peak and for loop instability. This is a design check, not a universal pass/fail threshold; the margin you need depends on the converter and the load.
A hypothetical bench sequence
The following is a labeled illustration, not a measured case. It is the sequence I would run before accepting a bead change as an emissions fix.
- Baseline the filter without the bead. Measure conducted emissions at the input terminals with the production-representative inductor and capacitor. Note the frequency of the highest quasi-peak reading and whether it sits near the filter’s calculated corner frequency f0 = 1/(2π√(LC)).
- Calculate the undamped corner and the damping factor. For a 5 V input filter with a 10 µH inductor and a 22 µF capacitor, f0 is roughly 10.7 kHz—low enough that the peak may sit below the conducted-emissions band of interest, but the same topology scaled to a higher corner can put the peak right in the scan. The damping factor depends on the source and load resistances and the capacitor ESR; if ζ is well below 1/√2, expect peaking.
- Add the bead and re-scan. Compare the peak frequency and amplitude. If the peak moved but did not shrink, the bead shifted the resonance rather than damping it. If a new peak appeared, the bead’s own reactance is likely interacting with the decoupling capacitor.
- Bias the bead at the actual operating current. If the vendor provides impedance versus frequency under DC bias, compare the biased curve to the 0 A curve at your frequency of interest. If the biased impedance is materially lower, the bead is doing less than the datasheet suggested.
- Check the bead’s self-resonance against the frequency of interest. Above self-resonance the bead behaves capacitively; if your emissions peak is above that frequency, the bead may be helping couple noise rather than blocking it.
- Damp the filter explicitly. TI’s note describes parallel and series damping networks and gives design equations for the damping resistor. A dedicated RC damping branch across the filter capacitor is a more predictable fix than hoping a bead provides the resistance.
Steps 4 and 5 require data the sources retrieved for this article did not provide. Treat them as evidence to gather from the specific part’s datasheet and vendor application notes, not as established numbers.
Where the prototype-to-production divergence enters
At 1k–10k units, the emissions failure usually does not come from a schematic change. It comes from a component decision made during production release:
- Bead substitution. A second-source bead with a similar nominal impedance at 100 MHz can have a different material, a different resistance-to-reactance ratio, and a different self-resonance. Murata’s note that impedance behavior varies by material is the reason a “same value” substitution is not the same filter.
- Capacitor dielectric change. A decoupling capacitor swapped from one dielectric to another changes ESR and the effective capacitance under bias, which moves the filter’s damping factor and corner frequency. The bead did not change; the network around it did.
- Inductor tolerance and saturation. The filter’s corner frequency scales with 1/√L. A production inductor at the low end of its tolerance, or one that saturates at the actual load current, shifts f0 and the peak.
- Placement and return path. A bead moved a few millimeters during a layout respin changes the parasitic inductance and capacitance of the network. The schematic is identical; the impedance profile is not.
None of these are exotic. They are the ordinary decisions that separate a prototype built by an engineer from a production build assembled by a contract manufacturer working from an approved vendor list. The forensic question is not “which bead is bad” but “which impedance changed, and at what frequency.”
What to put in the pre-production checklist
For a 1k–10k unit build, the checks that actually catch this failure mode are:
- Plot the filter output impedance and the converter input impedance on the same axes, across the frequency range of the emissions scan. Confirm separation per the Middlebrook criterion as described in TI SNVA538.
- Calculate the damping factor ζ from the actual source resistance, load resistance, inductor, capacitor, and capacitor ESR. If ζ is below 1/√2, add explicit damping rather than relying on a bead.
- Obtain the bead’s impedance versus frequency under the actual DC bias current, and compare it to the 0 A curve at the frequency of interest. If the vendor does not publish a biased curve, measure it or choose a part that does.
- Confirm the bead’s self-resonance is above the highest frequency you expect it to attenuate. Above self-resonance, the bead is a capacitor.
- Re-run the emissions scan after any bead, capacitor, or inductor substitution, including second-source approvals. A “form, fit, function” substitution is not an impedance substitution.
- Record the corner frequency and damping factor on the production drawing, not just the component values, so a future substitution can be checked against the filter behavior rather than the bill of materials.
FAQ
Does a ferrite bead always reduce emissions?
No. A bead changes the impedance profile of the network it sits in. If the surrounding LC filter is peaking, the bead can shift the peak or create a new one. The TI note’s description of undamped LC peaking is the mechanism; the bead is not exempt from it.
How do I know if my filter is peaking?
Calculate f0 = 1/(2π√(LC)) and the damping factor ζ from the source and load resistances and the capacitor ESR. If ζ is well below 1/√2, expect peaking near f0. Confirm with an impedance measurement or a conducted-emissions scan that shows a bump near that frequency.
What is the Middlebrook criterion in plain terms?
Keep the filter’s peak output impedance well below the converter’s input impedance across the relevant frequency range. TI SNVA538 states this as the condition to avoid oscillations; the note’s Figure 3 shows the two curves separated.
Why does the bead’s impedance matter under DC bias?
Ferrite material behavior changes with the DC current flowing through it, which can reduce the effective impedance at a given frequency. The sources retrieved for this article did not include a biased impedance curve, so verify against the specific part’s datasheet or vendor application note before assuming the 0 A curve applies.
What is self-resonance of a ferrite bead?
It is the frequency at which the bead’s parasitic capacitance resonates with its inductance, above which the bead behaves capacitively rather than inductively or resistively. The sources retrieved for this article did not include self-resonance data, so obtain it from the specific part’s datasheet and compare it to the frequency you need to attenuate.
Can a second-source bead cause a production emissions failure?
Yes, if the substitute has a different material, resistance-to-reactance ratio, or self-resonance. Murata’s guidance that impedance behavior varies by material is the reason a nominal impedance match at one frequency is not sufficient evidence of equivalence.
Sources
- Texas Instruments, Input Filter Design for Switching Power Supplies, SNVA538. https://www.ti.com/lit/an/snva538/snva538.pdf
- Murata, How to Select Ferrite Beads Considering the Characteristics of Digital Circuits. https://www.murata.com/products/emc/emifil/library/knowhow/basic/s2-chapter02-p1
