How to Read an LNA Datasheet: NF, P1dB and IP3 That Actually Matter

An LNA datasheet is three numbers wearing a lot of decoration: noise figure, gain, and linearity. Reading them together — through the lens of cascade analysis — is what separates a considered selection from shopping by headline.

Noise figure: the first stage owns your sensitivity

Friis' cascade formula has one dominant message: the noise contribution of every stage after the first is divided by the gain in front of it.

F_total = F₁ + (F₂ − 1)/G₁ + (F₃ − 1)/(G₁G₂) + …

Put a 0.9 dB-NF amplifier with 18 dB of gain in front of a 7 dB-NF receiver and the system noise figure lands near 1.1 dB. Put the same LNA after 3 dB of cable and filter loss and you get 3.9 dB — the loss adds directly. The practical rule: the LNA belongs at the antenna, and every tenth of a dB of connector and filter loss ahead of it is bought back at full price.

Datasheet fine print to check:

NF is specified at one frequency and temperature — 25 °C, band centre. Budget +0.1–0.2 dB at 85 °C.
NF at minimum gain is often much worse; if you plan to use gain-adjust, read that row.
"Typical" NF without a "maximum" column is a marketing figure. Our SL-LNA-2440 ships with the measured NF of that unit because distributions matter.

Gain: enough, and no more

Gain is the lever that suppresses second-stage noise — 15–20 dB is usually plenty. More gain is not free headroom; it drives the next stage harder and converts the linearity problem downstream into your problem.

P1dB and IP3: what survives strong signals

Sensitivity sells receivers; linearity keeps them working next to a Wi-Fi router. Two figures matter:

Output P1dB — where the amplifier runs out of volts. Keep expected peak output at least 6 dB below it.
Output IP3 — sets two-tone intermodulation. Each dB of blocker level above your plan costs 3 dB of intermod margin; a receiver on a shared ISM band should treat OIP3, not NF, as the spec most likely to fail in the field.

The tension is real: low NF and high IP3 fight for the same bias current. That is why "0.6 dB NF" parts with no IP3 spec deserve suspicion.

A five-minute vetting routine

Compute cascade NF with the LNA where it will actually sit.
Check NF and gain at your band edges and at 85 °C.
Put your worst-case blocker through the OIP3 maths.
Confirm stability (K > 1) below the band — bias networks oscillate at frequencies datasheet plots don't show.
Look for measured S-parameter files. If a vendor won't give you S2P data, they are telling you something.

Noise figure: the first stage owns your sensitivity

Friis' cascade formula has one dominant message: the noise contribution of every stage after the first is divided by the gain in front of it.

F_total = F₁ + (F₂ − 1)/G₁ + (F₃ − 1)/(G₁G₂) + …

Datasheet fine print to check:

NF is specified at one frequency and temperature — 25 °C, band centre. Budget +0.1–0.2 dB at 85 °C.

NF at minimum gain is often much worse; if you plan to use gain-adjust, read that row.

"Typical" NF without a "maximum" column is a marketing figure. Our SL-LNA-2440 ships with the measured NF of that unit because distributions matter.

P1dB and IP3: what survives strong signals

Sensitivity sells receivers; linearity keeps them working next to a Wi-Fi router. Two figures matter:

Output P1dB — where the amplifier runs out of volts. Keep expected peak output at least 6 dB below it.

Output IP3 — sets two-tone intermodulation. Each dB of blocker level above your plan costs 3 dB of intermod margin; a receiver on a shared ISM band should treat OIP3, not NF, as the spec most likely to fail in the field.

The tension is real: low NF and high IP3 fight for the same bias current. That is why "0.6 dB NF" parts with no IP3 spec deserve suspicion.

A five-minute vetting routine

Compute cascade NF with the LNA where it will actually sit.

Check NF and gain at your band edges and at 85 °C.

Put your worst-case blocker through the OIP3 maths.

Confirm stability (K > 1) below the band — bias networks oscillate at frequencies datasheet plots don't show.

Look for measured S-parameter files. If a vendor won't give you S2P data, they are telling you something.