RF·Tools

Link budget

A chain of stages collapses into four numbers: gain, noise figure, intercept point and compression. Sensitivity and dynamic range follow from those. But the most useful column is not the total — it is “noise share”: it tells you which stage is holding the noise figure, and improving any other one is wasted effort until that one is fixed.

VHF receiver LNA at the antenna Direct-sampling SDR

One stage per line, fields separated by a vertical bar: name | gain, dB | noise figure, dB | OIP3, dBm | P1dB, dBm. The last two are optional. A passive stage has a noise figure equal to its loss — leave it empty and that is what goes in.

Levels along the chain

-130-120-110-100-90-80-70Cable 10 m-2.0LNA+20.0Band-pass fil…-2.0Mixer-7.0IF filter-3.0IF amplifier+30.0signalnoiserf.akzami.com

The upper line is the signal, the lower one the noise; the gap between them is the signal-to-noise ratio. Under each stage is its gain in dB. You can see exactly where the ratio starts to melt away.

Stage by stage

Stage G, dB NF, dB ΣG, dB ΣNF, dB noise share signal, dBm SNR, dB
Cable 10 m -2.0 2.0 -2.0 2.00 36 % -112.0 21.2
LNA 20.0 1.0 18.0 3.00 25 % -92.0 20.2
Band-pass filter -2.0 2.0 16.0 3.02 1 % -94.0 20.2
Mixer -7.0 7.0 9.0 3.23 6 % -101.0 20.0
IF filter -3.0 3.0 6.0 3.48 8 % -104.0 19.7
IF amplifier 30.0 4.0 36.0 4.17 24 % -74.0 19.0

The noise share is the stage’s part of the Friis sum above the source noise. The column adds up to 100 %.

What you get

Gain36.0 dB
Noise figure4.17 dB
IIP3-11.3 dBm
OIP324.7 dBm
Input 1 dB compression-21.3 dBm
Sensitivity (noise in the band)-129.0 dBm
Dynamic range78 dB
Output signal-74.0 dBm
Output noise-93.0 dBm
Signal-to-noise ratio19.0 dB

Sensitivity is −174 dBm/Hz plus the bandwidth plus the noise figure: the level at which the signal equals the chain’s own noise.

the chain starts with “Cable 10 m” losing 2.0 dB, and that adds to the noise figure one for one. If it is a cable or a filter, an LNA belongs in front of it
some stages have no intercept point given — they are treated as perfectly linear, so the overall IP3 comes out better than reality
compression is summed with the same formula as IP3: that is an approximation, the real sum depends on phases, so leave a decibel or two of margin