L
Admin
Power pins of RF ICs need a path that looks low impedance at DC and high impedance at the frequencies where the pin would otherwise radiate or suck RF into the supply. Ferrite beads and wound RF chokes both do that job, but they fail in different ways.
Ferrite bead (chip bead)
- Lossy by design. It turns RF into heat, which is often what you want on a digital or mixed-signal rail.
- Impedance peaks in a limited band (often tens to hundreds of MHz) and then falls when parasitic C dominates.
- DC resistance and current rating matter: a bead that saturates at your bias current collapses to near zero ohms of RF impedance.
- Good for: VCO/PLL digital rails, baseband, Ethernet PHY, any rail that must kill harmonics without high Q.
RF choke / wirewound inductor
- Higher Q, less dissipative. It blocks RF more by reactance than by loss.
- Can resonate with decoupling capacitors and create supply peaking if you do not damp it.
- Better when you need low loss on a bias tee or an LNA/PA drain supply where every tenth of a dB counts.
- Watch SRF: above self-resonance the "choke" looks capacitive and stops helping.
Practical selection checklist
1. Plot Z(f) at your actual DC current (vendors sometimes only show small-signal curves).
2. Place a local bypass cap on the IC side of the bead/choke; place bulk caps on the supply side.
3. Check the anti-resonance between the choke and the caps with a quick simulation or by probing supply noise with a spectrum analyzer and a bias tee / coaxial probe.
4. For PA drains, prefer a choke (or printed bias line) sized so SRF is above the highest harmonic you care about, and add a series resistor or lossy bead if you see bias-network oscillation.
5. Never assume "higher ohm bead is always better" — a 1 kΩ bead that saturates at 50 mA is worse than a 300 Ω bead that stays linear.
Rule of thumb I use: beads on noisy digital / clock-adjacent rails; chokes (with damping) on RF bias where insertion loss and linearity matter.
Which bias networks have bitten you — bead saturation, or choke + cap peaking?
Ferrite bead (chip bead)
- Lossy by design. It turns RF into heat, which is often what you want on a digital or mixed-signal rail.
- Impedance peaks in a limited band (often tens to hundreds of MHz) and then falls when parasitic C dominates.
- DC resistance and current rating matter: a bead that saturates at your bias current collapses to near zero ohms of RF impedance.
- Good for: VCO/PLL digital rails, baseband, Ethernet PHY, any rail that must kill harmonics without high Q.
RF choke / wirewound inductor
- Higher Q, less dissipative. It blocks RF more by reactance than by loss.
- Can resonate with decoupling capacitors and create supply peaking if you do not damp it.
- Better when you need low loss on a bias tee or an LNA/PA drain supply where every tenth of a dB counts.
- Watch SRF: above self-resonance the "choke" looks capacitive and stops helping.
Practical selection checklist
1. Plot Z(f) at your actual DC current (vendors sometimes only show small-signal curves).
2. Place a local bypass cap on the IC side of the bead/choke; place bulk caps on the supply side.
3. Check the anti-resonance between the choke and the caps with a quick simulation or by probing supply noise with a spectrum analyzer and a bias tee / coaxial probe.
4. For PA drains, prefer a choke (or printed bias line) sized so SRF is above the highest harmonic you care about, and add a series resistor or lossy bead if you see bias-network oscillation.
5. Never assume "higher ohm bead is always better" — a 1 kΩ bead that saturates at 50 mA is worse than a 300 Ω bead that stays linear.
Rule of thumb I use: beads on noisy digital / clock-adjacent rails; chokes (with damping) on RF bias where insertion loss and linearity matter.
Which bias networks have bitten you — bead saturation, or choke + cap peaking?