L
Admin
[Technical article / FAQ, rfforum Editorial Team]
A PA that is stable and well-matched at room temperature can burst into oscillation or walk AM-AM after a cold soak — and the first blame is often the output match. Very often the culprit is a bias-network resonance that moved with temperature, ferrite, or electrolytic ESR.
1. Where bias resonances hide
- Drain / collector choke + bypass capacitors forming a high-Q tank near band or sub-harmonic
- Gate bias tee with a long high-Z trace acting as a stub
- Shared bias rails between stages that couple RF through the supply
- Ferrite beads whose impedance collapses or peaks differently at cold vs hot
2. Bench triage that separates match from bias
- Sweep S-probes or a spectrum sniff at the bias nodes (AC-coupled) while sweeping Pin — look for tones that track bias-network resonances, not the intended RF path
- Cold soak then re-check stability into the same load pull / cable VSWR used at room temp
- Temporarily damp the suspect node (series R or lossy bead) — if the "match" problem vanishes, you were chasing the wrong network
- Re-measure large-signal stability with the production bias layout, not a flying-lead lab supply
3. Design habit
Simulate or measure the bias network impedance seen by the RF transistor across frequency AND across temperature corners. Put a damping strategy in the first spin if the choke/bypass Q is high.
Discussion: on your last PA instability, was the root cause output match, bias resonance, or load VSWR — and do you require a cold-soak stability gate before tape-out?
— rfforum 编辑部 / Editorial Team
A PA that is stable and well-matched at room temperature can burst into oscillation or walk AM-AM after a cold soak — and the first blame is often the output match. Very often the culprit is a bias-network resonance that moved with temperature, ferrite, or electrolytic ESR.
1. Where bias resonances hide
- Drain / collector choke + bypass capacitors forming a high-Q tank near band or sub-harmonic
- Gate bias tee with a long high-Z trace acting as a stub
- Shared bias rails between stages that couple RF through the supply
- Ferrite beads whose impedance collapses or peaks differently at cold vs hot
2. Bench triage that separates match from bias
- Sweep S-probes or a spectrum sniff at the bias nodes (AC-coupled) while sweeping Pin — look for tones that track bias-network resonances, not the intended RF path
- Cold soak then re-check stability into the same load pull / cable VSWR used at room temp
- Temporarily damp the suspect node (series R or lossy bead) — if the "match" problem vanishes, you were chasing the wrong network
- Re-measure large-signal stability with the production bias layout, not a flying-lead lab supply
3. Design habit
Simulate or measure the bias network impedance seen by the RF transistor across frequency AND across temperature corners. Put a damping strategy in the first spin if the choke/bypass Q is high.
Discussion: on your last PA instability, was the root cause output match, bias resonance, or load VSWR — and do you require a cold-soak stability gate before tape-out?
— rfforum 编辑部 / Editorial Team
