L
Admin
[Technical article, rfforum Editorial Team]
Every LNA datasheet shows a Γopt (or Γs_opt) for minimum noise figure, and it is almost never S11*. Designing to Γopt improves NF; designing to S11* improves input return loss. You rarely get both without a cascode, feedback, or a balanced topology.
1. What Γopt is
Γopt is the source reflection coefficient that minimises F. It comes from the four noise parameters (Fmin, Rn, Γopt). Moving away from Γopt raises NF roughly with Rn·|Γs−Γopt|² / (stuff with (1−|Γs|²)). High Rn means the NF bowl is steep — you must hit Γopt carefully. Low Rn means you can trade match for almost no NF penalty.
2. The usual compromise
Handset and wideband LNAs often target a compromised Γs: NF within 0.2–0.5 dB of Fmin while keeping |S11| better than about −10 dB across the band. Pure Γopt match with |S11| ≈ −3 to −6 dB is fine behind an isolator or in a 50 Ω lab chain with a good driver; it is painful at an antenna port that must also meet VSWR specs.
3. Cascades and the next stage
Friis still rules: if the LNA gain is high, the next stage's NF matters less, so spending match budget on the input is rational. If gain is modest (first-stage GaAs with low |S21| at band edge), a slightly worse NF but flatter |S11| can make the cascade and the filter interaction more predictable.
4. Measurement sanity
Verify Fmin / Γopt on your own noise parameter setup or a trusted tuner; cheap EVB "NF" numbers often include board loss and a non-Γopt source. When you compare two matching networks, keep the same board loss de-embed and the same ENR correction.
5. When to stop chasing Fmin
- Antenna |S11| or filter interaction dominates the system NF more than 0.2 dB of LNA NF.
- Rn is low and you are already within a few tenths of Fmin.
- Stability circles say Γopt is too close to an unstable region — then back off.
Discussion: on your last sub-6 GHz LNA, how many tenths of a dB above Fmin did you accept to hold |S11| < −10 dB, and was Rn high or low on that device?
— rfforum 编辑部 / Editorial Team
Every LNA datasheet shows a Γopt (or Γs_opt) for minimum noise figure, and it is almost never S11*. Designing to Γopt improves NF; designing to S11* improves input return loss. You rarely get both without a cascode, feedback, or a balanced topology.
1. What Γopt is
Γopt is the source reflection coefficient that minimises F. It comes from the four noise parameters (Fmin, Rn, Γopt). Moving away from Γopt raises NF roughly with Rn·|Γs−Γopt|² / (stuff with (1−|Γs|²)). High Rn means the NF bowl is steep — you must hit Γopt carefully. Low Rn means you can trade match for almost no NF penalty.
2. The usual compromise
Handset and wideband LNAs often target a compromised Γs: NF within 0.2–0.5 dB of Fmin while keeping |S11| better than about −10 dB across the band. Pure Γopt match with |S11| ≈ −3 to −6 dB is fine behind an isolator or in a 50 Ω lab chain with a good driver; it is painful at an antenna port that must also meet VSWR specs.
3. Cascades and the next stage
Friis still rules: if the LNA gain is high, the next stage's NF matters less, so spending match budget on the input is rational. If gain is modest (first-stage GaAs with low |S21| at band edge), a slightly worse NF but flatter |S11| can make the cascade and the filter interaction more predictable.
4. Measurement sanity
Verify Fmin / Γopt on your own noise parameter setup or a trusted tuner; cheap EVB "NF" numbers often include board loss and a non-Γopt source. When you compare two matching networks, keep the same board loss de-embed and the same ENR correction.
5. When to stop chasing Fmin
- Antenna |S11| or filter interaction dominates the system NF more than 0.2 dB of LNA NF.
- Rn is low and you are already within a few tenths of Fmin.
- Stability circles say Γopt is too close to an unstable region — then back off.
Discussion: on your last sub-6 GHz LNA, how many tenths of a dB above Fmin did you accept to hold |S11| < −10 dB, and was Rn high or low on that device?
— rfforum 编辑部 / Editorial Team
