Noise Figure: When It Matters and When It Doesn’t
Noise Figure: When It Matters and When It Doesn’t
Noise figure tells us how much a network degrades signal-to-noise ratio under stated source and measurement conditions. It is valuable—but it cannot describe an installed receive antenna by itself.
At VHF and above, or behind a low-noise antenna and feed system, fractions of a decibel can matter. At lower HF in a noisy environment, chasing the smallest headline NF can waste dynamic range without improving what you can hear.
The Definition Is a Ratio
For a two-port network under stated conditions, the noise factor F is the ratio of input SNR to output SNR:
Noise factor: F = SNRin / SNRout
Noise figure: NF = 10 log10(F) dB
An ideal noiseless network has F = 1 and NF = 0 dB. A real device adds noise, so its output SNR is lower than its input SNR. An equivalent input noise temperature is often written, using the standard reference temperature T0 = 290 K:
Equivalent noise temperature: Te = T0(F − 1)
That conversion does not mean the device is physically at that temperature. It is another way to express its input-referred added noise.
Noise Figure Has Conditions
A defensible result states frequency, bandwidth, source impedance or reflection coefficient, device bias, gain state, temperature, calibration method and reference planes. A datasheet value measured in a matched 50-ohm system does not automatically describe a high-impedance probe, a balanced loop or a strongly reactive antenna terminal.
Active antennas combine the sensor and first stage. The incident field, antenna factor or effective height, sensor impedance, loss, protection network and amplifier noise all contribute to the field-referred result. Quoting the amplifier NF alone can hide the dominant boundary.
The First Stage Usually Matters Most
For cascaded stages with linear noise factors Fn and available power gains Gn, the Friis relationship is:
Cascade noise factor: Ftotal = F1 + (F2 − 1)/G1 + (F3 − 1)/(G1G2) + …
Enough low-noise gain before a lossy cable or noisy receiver can make later contributions small. But “enough” is the important word. More gain also reduces the blocker level required to reach compression, intermodulation or ADC clipping.
Compare Added Noise With Antenna Noise
A useful receive design compares the chain’s equivalent input noise with the noise actually delivered by the antenna in its environment. ITU-R P.372 shows that external radio noise varies strongly with frequency, location, time and environment. At HF, atmospheric and man-made noise can stand well above receiver noise; at a quiet site, on higher bands or with a low-gain sensor, receiver-added noise can become decisive.
If the antenna-plus-environment noise is already far above the following receiver noise, reducing NF further produces only a small change in total noise and SNR. If receiver noise is comparable, a lower system NF or more first-stage gain can materially help.
Linearity Can Be the Scarcer Resource
Broadcast transmitters, nearby amateur stations and wideband services can place far more total power at an active input than the wanted channel suggests. A beautifully low NF does not prevent compression, intermodulation, reciprocal mixing or digitizer overload.
Filtering helps only when it precedes the stage being protected. Attenuation can improve headroom and still leave the system externally noise limited. This is one reason a controlled attenuator test is so useful: if added attenuation lowers signal and noise together without reducing SNR, the system had noise margin available.
EchoTracer as an Active-Antenna Example
RF.Guru’s EchoTracer is an example of an active receive antenna in which the sensing element, protection, filtering, active interface and feed system must be considered together. The example does not assign a noise-figure number or guarantee an installed SNR. Those require a declared configuration, source condition, frequency, bandwidth, calibration plane and site-noise measurement.
A Measurement Sequence That Answers the Right Question
- Measure or bound the antenna-delivered noise with a calibrated receiver state and declared bandwidth.
- Insert known attenuation at the correct plane and observe whether wanted-signal SNR changes.
- Measure gain and NF across frequency under the actual source and bias conditions where practical.
- Record the strongest blockers and verify compression, intermodulation and ADC headroom.
- Include feedline loss, filters, bias-T loss and receiver NF in the cascade.
- Repeat after changes in location, antenna element, filtering or receiver settings.
The best front end is not the one with the smallest isolated NF. It is the one that adds little enough noise, supplies only the gain that is needed and remains linear in the real signal environment.
Primary and authoritative references
- ITU-R SM.1838-1 — Noise-figure measurement procedure
- ITU-R P.372-17 — Radio noise
- Keysight — High-accuracy noise-figure measurements
- Keysight — Noise-figure measurement by the Y-factor method
- ITU-R SM.1837-1 — Third-order intercept measurement
Mini-FAQ
- What does noise figure measure? It measures how much a network degrades SNR under stated source, frequency, bandwidth, bias, temperature and reference-plane conditions.
- Is lower NF always better at HF? Not materially when external antenna noise already dominates by a comfortable margin and strong-signal headroom is more limiting.
- Can amplifier NF describe a complete active antenna? No. Sensor impedance, antenna factor, loss, protection, gain, feed system and the incident noise field also matter.
- Why does the first stage matter most? Its added noise is not divided by preceding gain, while enough first-stage gain suppresses later stages’ input-referred contributions.
- Can more gain make reception worse? Yes. Excess gain can reduce blocker margin and cause compression, intermodulation or ADC clipping without improving SNR.
- What quick test reveals external-noise margin? Add known attenuation at a declared plane and compare wanted-signal SNR with fixed receiver settings.