What is the Friis noise formula?
F_total = F1 + (F2 − 1) ÷ G1 + (F3 − 1) ÷ (G1 × G2) + …, with every noise factor F and gain G as a linear power ratio. Convert the result back with NF = 10 log10(F).
Up to six stages go into this noise figure calculator, each with its gain and NF, and the Friis formula gives the total NF and gain of the chain. See how much each stage adds, along with the noise temperature, noise floor and receiver sensitivity.
A loss is a negative gain: a 2 dB filter is −2.
Passive part at 290 K: same as its loss. Blank pair = unused stage.
Optional: gives the noise floor and sensitivity.
At the demodulator. Blank is treated as 0 dB.
Optional: antenna temperature. Blank uses 290 K.
Cascaded noise figure
1.946 dB
Friis assumes matched stages, available power gains and noise figures measured with a 290 K source. Enter a loss as a negative gain. For a passive part at 290 K the noise figure equals its loss.
The floor uses kT0 = −173.98 dBm/Hz at 290 K, the exact value behind the rounded −174 dBm/Hz.
What each stage adds
| Stage | Gain | NF | Gain ahead | Adds to F | Share of added noise | Cascade NF |
|---|---|---|---|---|---|---|
| 1 | 20.0 dB | 1.0 dB | 0.0 dB | 1.26 | 45.8 % | 1.00 dB |
| 2 | −2.0 dB | 2.0 dB | 20.0 dB | 0.00585 | 1.0 % | 1.02 dB |
| 3 | −7.0 dB | 7.0 dB | 18.0 dB | 0.0636 | 11.3 % | 1.23 dB |
| 4 | 30.0 dB | 6.0 dB | 11.0 dB | 0.237 | 41.9 % | 1.95 dB |
Convert each stage to linear values, then apply the Friis formula: F = F1 + (F2 − 1) ÷ G1 + (F3 − 1) ÷ (G1 × G2) and so on. Two LNAs with 1 dB and 3 dB noise figures and 12 dB and 15 dB of gain, followed by an ADC with a 20 dB noise figure, give F = 1.519, which is 1.82 dB. The first stage dominates because later noise is divided by all the gain ahead of it.
Noise figure measures how much a stage degrades the signal-to-noise ratio. In a chain, each stage’s added noise is reduced by the gain that comes before it.
Noise temperature expresses the same quantity against the 290 K reference, and it cascades the same way:
The noise floor and sensitivity follow from the total noise figure:
The first term is −173.98 dBm/Hz, usually rounded to −174 dBm/Hz. If the source is not at 290 K, as with an antenna looking at cold sky, the floor is k (Ta + Te) BW instead.
LNA1 is 1 dB NF and 12 dB gain, LNA2 is 3 dB and 15 dB, and the ADC is 20 dB with 0 dB gain. The terms are 1.259 + 0.0628 + 0.1975, which sum to F = 1.519, or 1.82 dB. TI publishes 1.8 dB for this chain. The noise temperature is 150.6 K.
A chain of a 20 dB LNA with 1 dB NF, a 2 dB filter, a 7 dB loss mixer and a 30 dB IF amplifier with 6 dB NF has a noise figure of 1.95 dB. Put the filter ahead of the LNA and it becomes 3.61 dB, which is the 1.66 dB cost of moving the loss forward. These stage values are illustrative.
With the LNA first and a 1 MHz noise bandwidth, the noise floor is −173.98 + 60 + 1.95 = −112.0 dBm. A demodulator that needs 10 dB of SNR sets the sensitivity at −102.0 dBm.
Te = 290 × (F − 1).
| NF (dB) | F (linear) | Te (K) |
|---|---|---|
| 0.5 | 1.122 | 35.39 |
| 1.0 | 1.259 | 75.09 |
| 1.5 | 1.413 | 119.6 |
| 2.0 | 1.585 | 169.6 |
| 3.0 | 1.995 | 288.6 |
| 4.0 | 2.512 | 438.4 |
| 5.0 | 3.162 | 627.1 |
| 6.0 | 3.981 | 864.5 |
| 8.0 | 6.31 | 1540 |
| 10.0 | 10 | 2610 |
| 15.0 | 31.62 | 8881 |
| 20.0 | 100 | 28710 |
kT0 × BW × F at 290 K, in dBm, before any required SNR is added.
| Bandwidth | NF 0 dB | NF 3 dB | NF 6 dB | NF 10 dB |
|---|---|---|---|---|
| 1 Hz | −174.0 dBm | −171.0 dBm | −168.0 dBm | −164.0 dBm |
| 1 kHz | −144.0 dBm | −141.0 dBm | −138.0 dBm | −134.0 dBm |
| 12.5 kHz | −133.0 dBm | −130.0 dBm | −127.0 dBm | −123.0 dBm |
| 125 kHz | −123.0 dBm | −120.0 dBm | −117.0 dBm | −113.0 dBm |
| 200 kHz | −121.0 dBm | −118.0 dBm | −115.0 dBm | −111.0 dBm |
| 1 MHz | −114.0 dBm | −111.0 dBm | −108.0 dBm | −104.0 dBm |
| 10 MHz | −104.0 dBm | −101.0 dBm | −98.0 dBm | −94.0 dBm |
| 20 MHz | −101.0 dBm | −98.0 dBm | −95.0 dBm | −91.0 dBm |
| 100 MHz | −94.0 dBm | −91.0 dBm | −88.0 dBm | −84.0 dBm |
The Friis formula assumes each stage is matched to the next, uses available power gain and takes noise figures specified with a 290 K source. A mismatched interface changes both the real gain and the real noise figure, so use values measured in your impedance environment. Enter a passive loss as a negative gain with the same value as its noise figure, and remember that a hot cable or filter is noisier than its loss suggests.
An ADC can be treated as a last stage with 0 dB gain and its own noise figure, as TI does. Quote a mixer’s noise figure exactly as its data sheet specifies it, since single-sideband and double-sideband figures differ. And keep an eye on compression: the lowest noise figure is not useful if a strong nearby signal overloads the first stage, which is why many receivers accept a filter ahead of the LNA anyway.
F_total = F1 + (F2 − 1) ÷ G1 + (F3 − 1) ÷ (G1 × G2) + …, with every noise factor F and gain G as a linear power ratio. Convert the result back with NF = 10 log10(F).
F = input SNR ÷ output SNR with the source at 290 K, as a linear ratio, and NF = 10 log10(F) puts it in dB. A 3 dB noise figure is a noise factor of 1.995, and a noiseless stage has F = 1.
Yes. Alongside the cascaded noise figure it shows the total noise factor F as a linear ratio. Enter each stage’s gain and noise figure in dB.
Each later stage’s noise is divided by all the gain ahead of it. In TI’s example chain, a 3 dB second LNA behind 12 dB of gain raises the noise figure by only 0.2 dB.
No. Convert each noise figure and gain to a linear ratio first, apply the Friis formula, and convert the total back to dB. Gains can be added in dB, noise figures cannot.
For a passive part at 290 K the noise figure equals its loss. A cable with 3 dB of loss has a 3 dB noise figure, and if it sits ahead of the first amplifier that 3 dB adds directly to the system noise figure.
Te = 290 × (F − 1), with F the linear noise factor. A 3 dB noise figure is 288.6 K and a 1 dB noise figure is 75.09 K.
Sensitivity = −174 dBm + 10 log10(BW) + NF + the required SNR. With a 1.95 dB noise figure, 1 MHz of bandwidth and 10 dB of SNR that gives −102.0 dBm.
It is kT at the 290 K reference temperature, about 4 × 10⁻²¹ W/Hz. Calculated with the exact Boltzmann constant it is −173.98 dBm/Hz, which this calculator uses and which rounds to −174.
For noise figure, after: a filter ahead of the LNA adds its insertion loss directly. In the example below, moving a 2 dB filter ahead of the LNA changes the noise figure from 1.95 dB to 3.61 dB. Strong out-of-band signals can force the filter forward anyway, to keep the LNA out of compression.
See all free electronics tools →
ProtoFlow is free AI PCB design software: describe a circuit and get a schematic with real parts, PCB layout with an autorouter, SPICE simulation and Gerber files. Native KiCad, Altium and EAGLE import and export. Free on macOS, Windows and Linux.
Free to start. No card required.