Field Guide · term

Also known as: NF, noise factor, F

Noise figure (NF) is the amount, in decibels, by which a component or receiver worsens the signal-to-noise ratio of the signal passing through it — the ratio of SNR at the input to SNR at the output.1 Its linear form, the noise factor F, expresses the same thing as a plain ratio. Because every real device adds its own thermal noise on top of the noise already present, NF is always ≥ 0 dB (F ≥ 1); an ideal noiseless device has NF = 0 dB. Noise figure is the headline number that, together with bandwidth and required SNR, sets a receiver’s sensitivity.

LNA F1, G1 mixer F2, G2 IF amp F3, G3 F = F1 + (F2−1)/G1 + (F3−1)/(G1·G2) high G1 shrinks every later term — stage 1 dominates
Friis's cascade formula: a high-gain, low-noise first stage divides down every later stage's noise contribution, so the front end sets the system noise figure.

How it works

Feed a device a signal accompanied by exactly the standard thermal noise (kTB at 290 K). The device amplifies both, but it also injects its own noise, so the SNR at the output is worse than at the input. The noise factor captures that loss:

F = SNR_in / SNR_out and NF = 10·log₁₀ F

A perfect amplifier (F = 1, NF = 0 dB) would raise signal and noise equally and preserve SNR. A device with NF = 3 dB halves the SNR; NF = 10 dB divides it by ten.

The crucial insight for a receiver is that stages combine through the Friis formula for noise:

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

Each later stage’s noise contribution is divided by the gain that precedes it. If the first stage has high gain G₁, the (F₂ − 1)/G₁ term shrinks and every stage after it matters even less. This is why the first amplifier dominates the system noise figure — and why a good low-noise amplifier placed first, before any lossy cable or filter, is the single most effective way to lower a receiver’s noise figure.

Variants

  • Passive losses. A lossy component (feedline, attenuator, filter) has a noise figure equal to its loss: 3 dB of cable loss ahead of the LNA is 3 dB of noise figure that the LNA can no longer undo. Put the amplifier at the antenna, not at the radio, whenever possible.
  • Noise temperature. For very low-noise systems (satellite, radio astronomy), engineers use noise temperature T_e instead — it resolves fractions of a dB better than NF and adds linearly in a cascade. They are two ways of writing the same physics: NF = 10·log₁₀(1 + T_e/290).

Relevance to SDR

Noise figure is why an external LNA transforms a mediocre SDR. RTL-SDR dongles often have a noise figure of 5–8 dB or worse, degraded further by USB and clock spurs; placing a low-noise, high-gain preamplifier at the antenna makes the dongle’s own noise figure almost irrelevant, per Friis, and drops the effective system NF close to that of the LNA. The payoff is real weak-signal margin: lowering system NF by 6 dB is worth the same as a 6 dB better antenna or 4× the transmit power at the far end.

GopherTrunk itself is downstream of all of this — it processes the samples the front end delivers and cannot recover SNR the receiver’s noise figure has already thrown away. Its practical relevance to a GopherTrunk user is diagnostic: if a control channel decodes poorly and the signal is genuinely weak (low SNR at the ADC), the fix is upstream in the RF chain — a better LNA, shorter or better feedline, or an LNA moved to the mast — not a software setting.

Sources

  1. Noise figure — Wikipedia, definition of noise factor/figure and the Friis cascade formula. 

See also