Why is the LNA's noise figure so critical?

The Friis equation for cascaded noise figure is: NF_sys = NF_1 + (NF_2 minus 1)/G_1 + (NF_3 minus 1)/(G_1 times G_2). The first stage noise figure (NF_1) adds directly to the system NF. The second stage contribution is divided by the first stage gain (G_1). For an LNA with NF = 0.5 dB and G = 20 dB (100x): if the mixer has NF = 8 dB, its contribution is (6.31 minus 1)/100 = 0.053, or about 0.02 dB. The LNA dominates. If you replaced the LNA with a 3 dB NF amplifier, system NF goes from approximately 0.52 dB to approximately 3.02 dB, a 2.5 dB sensitivity loss. This means the receiver needs either 2.5 dB more transmitter power or 2.5 dB more antenna gain to achieve the same range. Conversely, anything before the LNA (cables, filters, switches) adds directly to the system NF. A 2 dB cable loss before a 0.5 dB NF LNA gives a system NF of approximately 2.5 dB. This is why the LNA should be as close to the antenna as possible.

What technologies are used for LNAs?

GaAs pHEMT (pseudomorphic High Electron Mobility Transistor): the workhorse for discrete LNAs from 1 to 100 GHz. NF of 0.3 to 0.5 dB at 2 GHz, 0.5 to 1.0 dB at 20 GHz, 1.5 to 2.5 dB at 60 GHz. Excellent gain and NF, moderate linearity. InP HEMT: the lowest noise technology available. NF of 0.1 dB at room temperature, approaching 3 Kelvin (0.04 dB) when cryogenically cooled. Used for radio astronomy and deep-space communications where ultimate sensitivity is required. Very expensive and fragile. SiGe BiCMOS: integrates LNA with mixer, VCO, and digital circuits on a single chip. NF of 0.5 to 1.5 dB at cellular frequencies. Lower performance than GaAs but much lower cost and higher integration. Dominates the cellular and Wi-Fi markets. CMOS: NF of 1.5 to 3 dB, lowest cost, highest integration. Used in consumer products where ultimate NF is not required. GaN: higher linearity (IIP3 10 to 20 dB higher than GaAs) at the cost of 1 to 2 dB higher NF. Used when strong interferers are present and dynamic range matters more than sensitivity.

How do you design an LNA?

LNA design requires simultaneous optimization of noise figure, gain, linearity, and stability. Noise matching: for minimum NF, the source impedance must equal Z_opt (the optimum noise impedance), which is generally NOT 50 ohms and NOT the conjugate match impedance. Presenting Z_opt to the transistor input gives NF_min but not maximum gain. The noise-gain tradeoff is visualized using noise circles on the Smith chart. The designer chooses a source impedance that gives acceptable NF (within 0.1 to 0.2 dB of NF_min) while maintaining adequate gain and input return loss. Stability: the LNA must be unconditionally stable (K greater than 1) at all frequencies, not just the design frequency. Out-of-band stability is ensured with resistive loading (at the cost of NF or gain) or by ensuring that gain at out-of-band frequencies is insufficient to sustain oscillation. Linearity: for most LNA applications, linearity is secondary to NF, but in dense signal environments (cellular base stations), high IIP3 is needed. Inductive source degeneration simultaneously provides noise matching, input matching, and improved linearity.

Receiver Front End

LNA

/ell-en-ay/ — Low Noise Amplifier

First receiver stage: gain 15-25 dB, NF 0.3-3 dB. Friis: NF_sys ≈ NF_LNA + (NF₂−1)/G_LNA. LNA dominates system NF. GaAs pHEMT: 0.3 dB @2GHz. InP: 0.1 dB (cryo: 3K). SiGe: 0.5-1.5 dB (integrated). Pre-LNA loss adds directly to NF. Noise matching: Z_opt ≠ Z₀ ≠ Z_conj. Inductive degeneration: NF + match + linearity.

GaAs: 0.3 dB

SiGe: 0.5-1.5 dB

Gain: 15-25 dB

Understanding LNAs

The LNA is the most critical component in any receiver. Its noise figure sets a floor on the system's ability to detect weak signals, and its gain determines how much the subsequent stages contribute to the total noise. Getting the LNA right is the single most impactful design decision in receiver architecture.

The fundamental challenge of LNA design is that minimum noise and maximum gain require different source impedances. The optimum noise impedance (Z_opt) is not 50Ω and is not the conjugate match. The designer must navigate this three-way tradeoff between noise, gain, and input match, using noise circles on the Smith chart to find the sweet spot.

LNA Equations

Friis cascade:
NF_sys = NF₁ + (NF₂−1)/G₁
+ (NF₃−1)/(G₁×G₂)
LNA: NF=0.5, G=20dB, Mixer NF=8dB:
NF_sys = 0.5 + 0.02 = 0.52 dB

Noise temperature:
T_e = T₀(F−1) = 290(10^NF/10−1)
NF=0.5dB: T_e=35K
NF=0.1dB: T_e=6.7K
Cryo InP: T_e=3-5K

Inductive degeneration:
Z_in = jω(L_g+L_s) + 1/(jωC_gs)
+ g_mL_s/C_gs

LNA Technology Comparison

Technology	NF @2GHz	Gain	IIP3	Application
GaAs pHEMT	0.3-0.5 dB	20-25 dB	−5 to +5	Satcom, BTS
InP HEMT	0.1-0.3 dB	25-30 dB	−10 to 0	Radio astro, DSN
SiGe BiCMOS	0.5-1.5 dB	15-20 dB	0 to +10	Cellular, Wi-Fi
CMOS	1.5-3 dB	10-15 dB	−5 to +5	Consumer IoT
GaN	1-2 dB	15-20 dB	+15 to +25	EW, dense env

Common Questions

Frequently Asked Questions

Why critical?

Friis: NF_sys ≈ NF_LNA (2nd stage divided by G_LNA). LNA NF 0.5+G20: mixer NF 8 contributes only 0.02 dB. Pre-LNA loss adds directly (2 dB cable = 2 dB system NF increase). LNA as close to antenna as possible. Every 1 dB NF = 1 dB less sensitivity = 25% less range.

Technologies?

GaAs pHEMT: 0.3 dB @2G, workhorse. InP: 0.1 dB, cryo 3K, radio astronomy. SiGe: 0.5-1.5 dB, highly integrated (SOC). CMOS: 1.5-3 dB, lowest cost, consumer. GaN: 1-2 dB NF but IIP3 +15-25 (dense/EW). Technology = NF vs cost vs integration tradeoff.

Design?

Z_opt (min NF) ≠ Z_conj (max gain) ≠ 50Ω. Noise circles on Smith chart: find sweet spot. Inductive source degeneration: simultaneous noise match + input match + linearity improvement. Stability: K>1 at all frequencies. Out-of-band: resistive loading if needed. Linearity vs NF tradeoff always present.

Related Terms

← LNA Bypass Load Pull →

Receiver Design

Request a Quote

Need LNA design, noise figure optimization, or receiver front-end development? Contact our team.

Get in Touch