This video challenges the common practice of blindly using ferrite beads for power supply noise filtering, demonstrating how they can create resonant circuits that increase noise and instability. Through practical measurements on test boards, the video explains that the typical pi filter (capacitor-ferrite-capacitor) forms a parallel LC resonator. This resonance can cause a significant impedance peak at the filter's output, leading to voltage ringing. It is shown that counterintuitively, increasing only the input-side capacitance can worsen this output impedance peak.
The solution presented is to damp the resonance by using sufficiently large capacitors on both sides of the ferrite. The video also highlights the critical impact of component selection, showing an unexpected case where an SMD capacitor performed significantly worse than a through-hole equivalent. For low-current applications, a simple resistor is proposed as a resonance-free alternative. Finally, the video warns about the often-overlooked effect of DC bias current, which can saturate a ferrite and reduce its impedance by up to 90%, and discusses measurement pitfalls like instrument crosstalk.
Rethinking Ferrite Beads: The Hidden Risks in Power Supply Design
0:00:01 The Ferrite Fallacy: Ferrite beads, commonly used to suppress power supply noise, can form resonant circuits with nearby capacitors, potentially increasing noise and causing instability.
0:01:55 The Pi Filter Problem: When used in a standard pi filter configuration (capacitor-ferrite-capacitor), the ferrite's inductance creates a parallel resonator. This can cause unwanted voltage ringing on the power rail.
0:03:59 Counterintuitive Behavior: Increasing the capacitance on the input side of the filter improves suppression from the source but can make the output impedance peak worse (5dB higher in the example), making the supply more unstable for the connected IC.
0:04:48 Understanding the Resonance: The pi filter acts like a voltage transformer at its resonant frequency. A large input capacitor effectively shorts one side, causing most of the resonant energy and highest impedance to appear at the output.
0:06:15 The Damping Solution: To prevent resonance, use large-value capacitors on both sides of the ferrite bead. This effectively damps the circuit (reduces the Q factor), resulting in a massive 30dB reduction in the impedance peak in the demonstrated test.
0:06:43 The Capacitor Surprise: Component choice is critical and can yield unexpected results. A test showed a standard SMD aluminum capacitor performing over three times worse (a >10dB difference) than a through-hole version, an issue similar in principle to problems that affected RTX30 series graphics cards.
0:07:54 A Resistor as a Safe Alternative: For low-power applications, replacing the ferrite bead with a small resistor (e.g., 10 ohms) can provide filtering without the risk of creating an inductive resonance, offering a more stable, "friendly" output impedance.
0:09:52 The Critical DC Bias Effect: A DC current flowing through a ferrite bead can saturate its core, drastically reducing its impedance—by up to 90% at only half its rated current. This severely degrades its filtering performance and is a crucial factor to check in the component's datasheet.
10:18 Accurate Measurement is Key: To reliably test these circuits, it's vital to calibrate for the frequency response of the measurement equipment and avoid issues like internal instrument crosstalk, which can significantly corrupt results.